Data Mining Reveals the Emotional Differences in Emails Written by Men and Women
Men and women include more anticipation words in workplace emails to members of the opposite sex, according to the first large-scale study of sentiment in electronic mail
Many behavioural psychologists believe that men and women use language in different ways. The conventional thinking is that women use language to foster personal relations whereas men aim for social position with a tendency to be more confrontational.
But evidence is difficult to come by because of the challenge in gathering and objectively analysing large bodies of language both spoken and written.
All that has changed in recent years thanks to the new science of sentiment analysis. This relies on the creation of vast databases in which words are marked as either positive or negative and associated with one of the eight fundamental emotions: joy, trust, fear, surprise, sadness, disgust, anger and anticipation.
It is then a relatively simple matter to data mine a corpus of digital text to see which emotions dominate. A growing number of companies are using this kind of analysis to monitor the emotions associated with tweets about their products, for example.
Today, Saif Mohammad and Tony Yang at the Institute for Information Technology in Ottawa, Canada, analyse some 200,000 workplace emails sent between October 1998 and June 2002 by 150 people in senior managerial positions.
The question they ask is whether the language in emails sent by women differs significantly from the language used by men. The answer provides some interesting insights into gender dynamics in the workplace.
(Continued below…)
This Week’s Top 5 Posts
The True Size of the Shadow Banking System Revealed (Spoiler: Humongous)
Voyager 1 May Be Caught Inside an Interstellar Flux Transfer Event
Text Analyser Reveals Emotional Temperature of Novels and Fairy Tales
Wealth in Africa Mapped Using Mobile Phone Data
“Ballooning” Spiders Use Electrostatic Forces To Generate Lift
The database chosen by Mohammad and Yang is the Enron email corpus because it is the only large publicly available collection of emails. The messages it contains are mostly about official business but there are also some personal communications.
Mohammad and Yang filtered the data set by removing emails with fewer than 50 words and more than 200 words. They then studied the name of each sender and identified 89 male and 41 female correspondents. (They removed the emails from the other 20 correspondents whose gender they were unable to determine.)
That left 32,000 emails, 20,000 of them from men and 12,000 from women. “We then determined the number of emotion words in emails written by men, in emails written by women, in emails written by men to women, men to men, women to men, and women to women,” they say.
The results are revealing. Mohammad and Yang’s main conclusions are these:
· When writing to women, both men and women use more joyous and cheerful words than when writing to men
· Both men and women use lots of trust words when writing to men
· Women use more cheerful words in emails than men.
· Women tend to share their worries with other women more often than men with other men, men with women, and women with men
· Men prefer to use a lot of fear words, especially when communicating with other men
· Both men and women are far more likely to use anticipation words when emailing a member of the opposite sex than in same-sex communication.
That provides a unique insight into the nature of communication between men and women in the workplace. And Mohammad and Yang want to go further. These guys are developing a Google app that will allow users to track their emotions towards the people they correspond with in Gmail. They plan to make a public call for volunteers willing to share their data for research purposes.
That sounds like fun. Interested parties should keep an eye out for the announcement and include plenty of joy, trust and anticipation words in their reply ;)
Ref: arxiv.org/abs/1309.6347 : Tracking Sentiment in Mail: How Genders Differ on Emotional Axes
