Slashdot Mirror
Data Mining Reveals the Emotional Differences In Emails From Men and Women
KentuckyFC writes "Sentiment analysis relies on vast databases of common words which are marked as positive, negative or neutral and associated with one of the eight fundamental emotions: joy, trust, fear, surprise, sadness, discuss, anger and anticipation. It is then a straightforward matter to search Tweets, novels and even fairy tales to see what emotions appear. Now, researchers have carried out the first large-scale study of sentiment in workplace emails. They examined the emotions associated with words in over 30,000 emails and analyzed the emotional differences between messages sent by men and women. It turns out that women use more cheerful words in emails than men, that men use more fear words, especially when communicating with other men, and that both men and women are far more likely to use anticipation words when emailing a member of the opposite sex. The same researchers say they are developing a Google app that will allow users to track their own emotions towards the people they correspond with in Gmail. And they plan to make a public call for volunteers willing to share their data for research purposes."
1) Men's emails are 90% more likely to include the word "balls" than women's.
2) Men are 74% less likely to ever email someone a link to an Oprah Magazine story.
3) No man has ever used the phrase "K, Luv Ya!" in an email, ever.
The cow says "Moo." The dog says "Woof." The Timothy says "Thanks, valued customer. We appreciate your input."
I'm not sure I ever felt discussing, though since I'm discussing it, I guess... I am discussed.
I'm god, but it's a bit of a drag really...
Where are "snark" and "sarcasm" found on the emotions list?
The Internet suggests those should be 90% of the total.
nah thats already pretty well documented
fuck you
second, hey they choose to use google...
i think people probably already have a good idea about what they actually think about the other people that they email back and forth with. a google app is just going to tell people things that they already know about themselves.
Admittedly, I only skimmed both the article and the study, but I did not see a link to the list of words and their emotional associations. This could be useful if they'd let us at it.
Of course men and women use different language in their emails. Young men would use different language than middle aged or older people do. A person emailing a friend would have different language than when they email their boss. This is not indicative of there "emotions". This is indicative of their education, wisdom, and who they are having a conversation with and the topics of discussion.
This whole article discusses work done on an absolutely false premise. Emails can not be used to determine your emotional state, any more than tweets can not be used to determine your psychological state.
Pack it in you eugenics morons! We are on to your game!
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Email language also vary based on where you're from, what your history is, your income level, your education level, how many fingers you've lost to chainsaws...
I think Slashdot had an article a while back about being able to use your email language to track down individual people because there are so many different variables.
Nothing new there then!
Fascinating!
Let's try the same thing with song lyrics:
Mama Cass:
Stars shining bright above you
Night breezes seem to whisper "I love you"
Birds singing in the sycamore tree
Dream a little dream of me
Say "Night-ie night" and kiss me
Just hold me tight and tell me you'll miss me
While I'm alone and blue as can be
Dream a little dream of me
Stars fading but I linger on, dear
Still craving your kiss
I'm longing to linger till dawn, dear
Just saying this
Sweet dreams till sunbeams find you
Sweet dreams that leave all worries behind you
But in your dreams whatever they be
Dream a little dream of me
And DJ Assault:
(2x)
Ass, titties, ass 'n titties
Ass ass titties titties, ass 'n titties
(2x)
Big booty bitches thats where it gets
Come on, hoe, let's go to the easy rest
When I see ass, titties, ass 'n titties
Ass, ass, titties, titties, ass 'n titties.
(4x)
Ass, ass, ass, ass
(2x)
If you a light-skinned bitch that think you the shit,
I can buy you, hoe, 'cause bitch I'm rich.
I see broke-ass hoes, broke-ass hoes,
Broke-ass hoes, broke-ass hoes.
(4x)
Hoes, hoes, hoes, hoes
(2x)
If you a freaky-dancin' ho, keep shakin' that shit
Let's see how you shake it on top of my dick
And you'll say "Assault, I'm cuming. Assault, I'm cuming.
"Assault, I'm cuming. Assault, I'm cuming."
(4x)
Cumin', Cumin', Cumin', Cumin'
(2x)
Stankin ass bitches that need to wash up
Dont get mad when i dont want to fuck
You need soap and water, soap and water
Soap and water, Soap and Water
(4x)
Water, Water, Water, Water
The difference is subtle, but in my opinion quite discernible.
The article stated nothing about anyone's "emotional state". Read TFA, digest TFA, comprehend TFA.
Males and females have absolutely no difference between them, despite the overwhelming experiential, physiological, neurological, anatomical, and hormonal evidence to the contrary, you sexist pigs, and how dare you even conduct this research into the issue!
Now if you'll excuse me, I need to go open a jar of mayo for the little lady, though that in no way suggests she lacks the wrist and forearm strength to grasp and twist just as well as any man.
Even the most basic data mining can reveal a lot about your hopes and dreams.
"eight fundamental emotions: joy, trust, fear, surprise, sadness, discuss, anger and anticipation"
Great. When my wife tells me I need to be more open about my feelings I just need to discuss the relative merits of procedural and nonprocedural programming languages
I second that emotion.
--Smokey
It is then a straightforward matter to search Tweets, novels and even fairy tales to see what emotions appear. Now, researchers have carried out the first large-scale study of sentiment in workplace emails. They examined the emotions associated with words in over 30,000 emails and analyzed the emotional differences between messages sent by men and women.
If you are determining that someone is of an emotion, you are determining their emotional state. If you are determining an emotional difference, you must know two states of emotion. Logic and reason is not very complex, but many people choose not to use either.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
I can hardly wait for the full science to come out.
I anticipate having a gigundamous brain orgasm when I process the data.
It's what I live for; it's what I write e-mails for; it's what I read comments for but then feel ashamed and disgusted.
I think that's a bunch of FUD.
I find TFS very discussing, you insensitive clod!
If you are determining that someone is of an emotion, you are determining their emotional state. If you are determining an emotional difference, you must know two states of emotion. Logic and reason is not very complex, but many people choose not to use either.
Slow down. Read it again, it says "emotions associated with words" and "emotional differences between messages". That says nothing about emotional state. Nor does it imply those are the true emotions of the authors. This is about how men and women use language differently. Why you read so much more into that, I don't know.
Even the journal article linked to by the blog seems to be scant on details. I was hoping there'd at least be a few appendices on these things, but no such luck. Anyone know exactly what "joy words," "fear words," et al are?
Thanks in advance.
It is apparent that you missed "The same researchers say they are developing a Google app that will allow users to track their own emotions towards the people they correspond with in Gmail."
If you choose to ignore what the article claims that is not my issue, but yours. I didn't read anything into the article, I read the whole article.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
This research is as valuable as the worth of a beautiful gold button which has been determined to be brass.
I have serious doubts that any program could determine the actual "emotional" meaning of that phrase.
The problem with all research of this nature is that human language is filled with reverses of logic and emotion in order to convey more subtle states of thought. Not to mention, as a writer I frequently change to a feminine voice for either a specific third person effect or simply dialogue for a female character. "Not to mention" being an example of reversal in logic.
Colloquialism is another killer. "I don't care to" carries an entirely different meaning in different parts of the US.
And yes, I will do that in emails if the recipient knows me.
Of course men and women use different language in their emails. Young men would use different language than middle aged or older people do. A person emailing a friend would have different language than when they email their boss. This is not indicative of there "emotions". This is indicative of their education, wisdom, and who they are having a conversation with and the topics of discussion.
I don't the researchers claim that the words directly reflect the writers emotions. When people write persuasively they instead try to stir emotions in their readers. For example, if I write: "I am deaply concerned that our current strategy exposes us to unnecessary risk." I am fear words in the hope that my readers will take what I have to say more seriously.
You could just as easily use those words due to ignorance and be repeating what you heard someone else say, and not intentionally using fear correct? How would a person or application be able to discern the difference? It can't because, as with any language, there is a tremendous amount of subjectivity in the words.
To take this a bit further, we often apply emotional words in an exactly different theme than our emotion. Take satire as a massive example. Another smaller example would be the person who just experienced a death of a close loved one that realizes they are emotionally distraught so intentionally choose words to mask their emotion (or for someone with knowledge of psychology, to facilitate changing their emotional state).
These small examples show that the claim is based on the false premise that you can know (via empathy, psychological profiling, etc..) what someone means when communicating. This is counter intuitive with even a basic understanding of language. One of the reasons we have so many different languages is that we have never perfected one that expresses our precise meaning, let alone our psychological reference for making our statements.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Are you one of the many people Ive come across that think the Internet isn't real? Meaning what you say and do to others on the Internet cant hurt someone else feeling and emotions because well its the Internet?
Jack of all trades,master of none
Seriously, do I give any hint that communication in any medium is different (internet vs. diary vs. book)? The point I raise is not that you can't harm someone in a communication, the point is that you can't know their intent by the words chosen from a single email.
If I was to analyze a threaded communication, I would get an idea that it's possible your "piss off" email was a joke and not malicious. Having no such thread and looking at the "piss off" email I can make no such speculation rationally, and should make no claim that it's possible. Especially considering that when I read the thread and assume you were joking, I may be very very wrong.
The point is not that communication is not real depending on the medium. The point is that we can't use a single communication as a reference to claim to know a person's emotional state or psychological state.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Elaine dumped her boyfriend because he would not use exclamation points when they might apply.
She in turn was fired for using them inappropriately (she was an editor for a publisher).
It is another illustration of the emotional differences between men and women and how they communicate
Wouldn't have asked if i thought any different. Which makes your point.
Jack of all trades,master of none
Emails can not be used to determine your emotional state
"Listen, you goddamned son of a bitch, one more fuckup and I'll pull your fucking teeth out with a goddamned rusty pair of pliars!"
Ya think?
No room for your facts here. Only propaganda.
Exactly what language give you the idea that I claim that a communication could not cause harm? I never make that claim, and make no implication. I claim that communications are subjective.
This is not indicative of there "emotions". This is indicative of their education, wisdom, and who they are having a conversation with and the topics of discussion.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Come now mods, this is absolutely not a troll.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Pack it in you eugenics morons! We are on to your game!
I sense irritation and a vague sense of intellectual superiority...
Woah... James Randi, here I come!
If it's for-profit but free, you're not the customer -- you're the product (e.g., the Slashdot Beta's "audience").
Irritation? I think it's worse than that with the eugenics people. I side with Socrates over the Sophists, and despise this type of rhetoric for the same reasons. Philosophy should not be used for personal gain, but should be used for bettering society. I may actually be irritated because people are repeatedly being duped by the same propaganda, and on a site where people are supposed to be intellectuals it's befuddling that the same rhetorical tricks seem to always work.
In the last year, there have been at least 5 different articles of "science" claiming to be able to work magic with words. This is everything from labeling you as a psychopath or a sociopath by your 1024 or less character tweets, to determining if you will be a future criminal by your DNA, to this one claiming to be able to know your emotion by a single email (or the emotions involved while you created your email). None of this is possible, but since they wrap it in a box labelled "science" people seem to ignore the content.
Sense of intellectual superiority? That is an entirely subjective, quite possibly stated as an ad hominem.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Worth pointing out that all the differences reported were pretty small. For example, the largest difference I found in all the tables was in the use of trust words, and the largest difference quoted in that was 4.3% (if I read the small numbers right, but in any case something slightly over 4%). All the other differences were smaller than that. So I think the discussion here makes the frequent error of magnifying the differences.
How convenient. No redundancy.
It's kind of retrograde, limited, and descriptively inadequate to insist on defining men and women first by gender. An alternative would be to cluster people by using a common vocabulary. You'll still get lots of gender stuff coming out but you can reflect the fact that there are lots of different ways to "do" gender. This is something we all know--there's not just one type of guy, nor one type of woman and gender--while often relevant--isn't always.
Here's a paper on the matter. "Computational sociolinguistics": http://arxiv.org/abs/1210.4567
The abstract:
We present a study of the relationship between gender, linguistic style, and social networks, using a novel corpus of 14,000 users of Twitter. Prior quantitative work on gender often treats this social variable as a binary; we argue for a more nuanced approach. By clustering Twitter feeds, we find a range of styles and interests that reflects the multifaceted interaction between gender and language. Some styles mirror the aggregated language-gender statistics, while others contradict them. Next, we investigate individuals whose language better matches the other gender. We find that such individuals have social networks that include significantly more individuals from the other gender, and that in general, social network homophily is correlated with the use of same-gender language markers. Pairing computational methods and social theory thus offers a new perspective on how gender emerges as individuals position themselves relative to audiences, topics, and mainstream gender norms.