Ranking authors by their adverb use

The famously spare Hemingway used 80 words ending in -ly per 10,000 words of prose; JK Rowling uses 140 adverbs per 10,000 words, and EL James uses 155. Read the rest

The word "sheeple" is now in the dictionary, with Apple fans as example

Merriam-Webster added "sheeple" to their dictionary. It's defined as "people who are docile, compliant, or easily influenced : people likened to sheep." Here's one of the two usage examples they include:

Apple's debuted a battery case for the juice-sucking iPhone—an ungainly lumpy case the sheeple will happily shell out $99 for.
Read the rest

Train your AI with the world's largest data-set of sarcasm, courtesy of redditors' self-tagging

Redditors' convention of tagging their sarcastic remarks is a dream come true for machine learning researchers hoping to teach computers to recognize and/or generate sarcasm. Read the rest

Tracing how languages are related to one another

In this video for Mental Floss, linguist Arika Okrent and illustrator Sean O’Neill share the keys to understanding commonalities between languages.

[via Laughing Squid] Read the rest

Gendered verbs charted over 100,000 stories

Data scientist David Robinson tracked the proximity of verbs to gender across 100,000 stories. She screams, cries and rejects. He kidnaps, rescues and beats.

I think this paints a somewhat dark picture of gender roles within typical story plots. Women are more likely to be in the role of victims- “she screams”, “she cries”, or “she pleads.” Men tend to be the aggressor: “he kidnaps” or “he beats”. Not all male-oriented terms are negative- many, like “he saves”/”he rescues” are distinctly positive- but almost all are active rather than receptive.

The chart on which types of violence are associated with men and women is predictable stuff: poison from the ladies, beatings from the gentlemen.

It follows on from Julia Silge's look at gender roles and text mining. See also an attempt to do likewise with plot arcs. Read the rest

How Disney uses language in its music

The Sideways YouTube channel digs into the many different ways Disney has used foreign languages in its animated musical movies over the years. Read the rest

What the Terminator says instead of "Hasta La Vista, baby" in the Spanish version of Terminator 2

In the Japanese version, he says "Cheerio then, love." Read the rest

How do new words get in the dictionary?

Kory Stamper, author of the new book Word by Word: The Secret Life of Dictionaries describes three criteria Merriam-Webster uses for inclusion of words like truther, binge-watch, photobomb and the 1,000 other words that make the cut in a typical year. Read the rest

Learn 12 different accents in under four minutes

Dialect coach Sammi Grant gives a crash course in a dozen different accents including my favorite, the Transatlantic accent. (Think Audrey Hepburn and Cary Grant. More on that here.)

"I’m legally blind and one of the reasons I got into dialect coaching is because I love to hear people’s voices and help people find the range of their voices," Grant says.

(via Laughing Squid)

Read the rest

The linguistic backflips used by Deliveroo to pretend its employees are independent contractors

Deliveroo is a "gig economy" company that hires people to cycle around big cities, delivering meals, while pretending that all their riders are actually "independent contractors" running their own businesses through which they subcontract to Deliveroo, thus dodging any need to pay benefits or comply with basic labor, health and safety rules. Read the rest

Your word of the day

Read the rest

AP stylebook now allows the "singular they" in some instances

The use of "them" as a gender-neutral pronoun goes back hundreds of years, but the weaponizing of grammar as a way to tell people how to talk (rather than a way to understand what speakers are saying) made the practice anathema, creating strong headwinds for people looking to adapt usage to accommodate a spectrum of gender identities. Read the rest

Deflationary Intelligence: in 2017, everything is "AI"

Ian Bogost (previously) describes the "deflationary" use of "artificial intelligence" to describe the most trivial computer science innovations and software-enabled products, from Facebook's suicide detection "AI" (a trivial word-search program that alerts humans) to the chatbots that are billed as steps away from passing a Turing test, but which are little more than glorified phone trees, and on whom 40% of humans give up after a single conversational volley. Read the rest

A Gmail plugin to do "emotional labor"

Install the Emotional Labor extension and it will automatically add social niceties to your outbound mail -- phrases like "Hey, Lovely! I've been thinking of you." (via 4 Short Links) Read the rest

How to talk about shitgibbons with your high school English class

The introduction of "shitgibbon" into America's political discourse is a teachable moment for English classes studying the evolution of language. Read the rest

How to talk about Trump using trumpspeak (and why you should)

"Impeach the guy. He’s an unstable person. Yemen? Disaster. Everyone knows it’s true. Even he knows it. Awful. Everyone agrees." Read the rest

How "futureless" languages impact political thought

There are certain languages that don't differentiate between the present and the future. Estonian is the classic example of a "futureless tongue." According to new research by Efrén O. Pérez, co-director of Vanderbilt University's Research on Individuals, Politics & Society Lab and Margit Tavits, professor of political science at Washington University, language has a sizable impact on how we think about future-oriented policies. As William S. Burroughs said, language is a virus. From their scientific paper in the American Journal of Political Science:

Can the way we speak affect the way we perceive time and think about politics? Languages vary by how much they require speakers to grammatically encode temporal differences. Futureless tongues (e.g., Estonian) do not oblige speakers to distinguish between the present and future tense, whereas futured tongues do (e.g., Russian). By grammatically conflating “today” and “tomorrow,” we hypothesize that speakers of futureless tongues will view the future as temporally closer to the present, causing them to discount the future less and support future-oriented policies more. Using an original survey experiment that randomly assigned the interview language to Estonian/Russian bilinguals, we find support for this proposition and document the absence of this language effect when a policy has no obvious time referent. We then replicate and extend our principal result through a cross-national analysis of survey data. Our results imply that language may have significant consequences for mass opinion.

Read the rest

More posts