The 2018 Olympic Winter Games in Pyeongchang, South Korea, are a marketing opportunity for pretty much every major company due to the swarms of viewers plugging in and streaming.
It’s also been a good time for Mike Elchik’s Carnegie Mellon University spinoff, WeSpeke, to announce its largest partnership to date — a deal with CNN.
WeSpeke — which Mr. Elchik describes as a hybrid of Match.com, Facebook, the Post-Gazette and Skype — is a language learning website and app meant to connect users by their common interests in order to spur conversation.
“There are people just like you in France and China with the same interests as you,” Mr. Elchik said. “You can use math to narrow everyone down to people very much like you.”
In late January, WeSpeke and CNN announced a partnership for a new platform called en.news, a free website and mobile app that automates English lessons from existing editorial content on CNN.com.
So a story about U.S. snowboarding gold medalist Shaun White is now instant English practice.
“You’re not going to change your habit of consuming the news,” Mr. Elchik said. “Everybody tends to get up and check the sports scores or headlines ... we’re allowing you to follow your habits and learn English at the same time.”
The founder and CEO calls his Green Tree-based startup a “disruptive” entrant into the digital English marketplace.
The firm, which employs 15 to 20 workers, has secured patents related to its expansive social network and matching technology, and it’s awaiting another for its natural language processing software — a branch of artificial intelligence that helps computers to understand human language.
The technology was largely developed through research by WeSpeke co-founder Jaime Carbonell, who is also director of CMU’s Language Technologies Institute in Oakland.
WeSpeke matches users learning new languages based on various factors, including their native tongue, the language they’d like to learn, their proficiency and their interests.
According to a patent filing for the company’s matching technology, the WeSpeke system uses two key metrics from user profiles to determine which learners are compatible with one another.
A screenshot of WeSpeke's patent for user-matching technology. (U.S. Patent & Trademark Office)
“Point match scores,” measure the strength of a single shared interest between two users and “compatibility match scores” are an aggregate measure of multiple shared interests. The system suggests a few friends to connect with, and a user may then engage in a conversation through text-based messaging or video calling.
Mr. Elchik did not comment on the number of active users on the platform, but said there are “millions of exchanges every month.”
Those conversations help drive a more natural type of practice.
“Learning a language is inherently human in communication,” Mr. Elchik said. “If you’re the only person in the world, there’s no such thing as language because there’s nobody to hear you and speak.”
WeSpeke’s en.news partnership with CNN relies on patent-pending natural language processing software that creates lessons from text-based content, like articles.
Tapping into CNN’s newsfeed, the startup can determine which stories are trending and automate them into lessons.
Articles are deconstructed into parts of language, such as nouns, verbs and adjectives and then transformed into a gamified lesson.
en.news, a partnership between WeSpeke and CNN, uses proprietary natural language processing technology to create English language lessons from CNN content. (Courtney Linder/Post-Gazette)
Repurposing original reporting to teach language learners is a way for CNN to increase engagement and for WeSpeke to leverage the national media company’s brand.
Mr. Elchik said there is potential for CNN to expand its audience from mostly those in their 40s or 50s, to a younger demographic of English language learners who are typically 18 to 28.
The partnership will also augment CNN and WeSpeke’s existing business models, which both rely on advertising, Mr. Elchik said. He did not disclose further financials about the companies’ relationship.
WeSpeke has raised at least $6.2 million in seed and venture capital rounds, according to Crunchbase, a portal for private companies to self-report some or all of their funding.
The Language Technologies Institute’s work, tied to a renowned linguistics department just down the road at the University of Pittsburgh — not to mention the commercial success of East Liberty-based language learning app Duolingo — are all part of a Pittsburgh narrative that is shaping the future of language technology.
Mr. Elchik should know. He speaks six languages and has traveled about 2.5 million miles in his lifetime.
“It’s really important as you move around the world that you understand the perspective, the culture and the language of the people who you’re meeting with,” he said.
“It’s so much more interesting to be a citizen of the world in that regard and not try to force everyone onto your platform.”
Courtney Linder: clinder@post-gazette.com or 412-263-1707. Twitter: @LinderPG.
Join the Tech.pgh Community!
Sign up for Courtney Linder's free newsletter.
First Published: February 14, 2018, 12:06 p.m.
