That’s what she said.
The delightful little phrase that highlights the previous speaker’s (usually) unintentional double entendre. TWSS jokes are a staple of the young person’s vernacular (and that of NBC writers).
The double entendre as a figure of a speech has been around for ages (Shakespeare’s plays are absolutely loaded with them, such as this gem from one of my favorite plays of his, Twelfth Night, in which a character’s hair is described as such: “…it hangs like flax on a distaff; and I hope to see a housewife take thee between her legs and spin it off”), and though the notoriously vulgar tone of most such phrases tend to paint these types of euphemisms as cheap and immature, they actually require a delicate understanding of language (particularly the multiple meanings of certain words and phrases).
It’s not as easy to understand euphemisms and double entendres as most people believe. For example, despite having a ready grasp on a second language’s grammar and vocabulary, it is extremely easy to miss subtle nuances when interacting with native speakers. Stories of unintentional double entendres/euphemisms in such situations are common.
With such a learning curve for this type of expression, it’s hard to imagine, say, a computer ever being able to grasp it. I mean, robots couldn’t have a translatable-to-humans sense of humor. That’s just preposterous:
And yet, in what is either the biggest waste of science and tech ever or is the most ingenious use of it, University of Washington computer sciences have developed a program that can recognize TWSS jokes.
You heard me right, galleons.
To achieve this, our scientists began by studying two texts, one containing 1.5 million erotic sentences and another with 57,000 from standard literature. They then evaluated nouns, adjectives, and verbs with a “sexiness” function. Words with a high sexiness function are prime TWSS fodder (examples would be things like ‘rod,’ ‘meat,’ ‘hot,’ and ‘wet’- those with multiple meanings).
The program they created, DEviaNT (Double Entendre via Noun Transfer), looks for those words with a “high sexiness function” and was trained by gathering jokes from twssstories.com and non-humorous text from sites like Wikiquote. DEviaNT has the difficult task of identifying, from this large sampling of text, sentences that contain potential euphemisms and follow a particular structure.
DEviaNT was about 70% accurate, but scientists believe that a more even data set for training purposes could result in a precision rate of 99.5%.
So… when the robot uprising finally happens, our mechanical masters will at least be able to make dirty jokes while they slaughter and subjugate us.
At least that’s something, right?