Software That Speaks Human: Meet Contextors, a Company That Takes the Guesswork out of Language Learning (Podcast)
This past month, we had the opportunity to sit down with a number of Israeli edtech companies at a local event. One of the companies was Contextors, a company focused on “perfecting the way computers use natural language.” This might seem like a pedantic pursuit, but some of their products could revolutionize the way we communicate through machines.
For example, one of the Contextors products is a dictionary app called Literāte. Literāte provides definitions within ebooks to help contextualize words with multiple meanings. Take Harry Potter, for example. While the word “potter” might mean “a person who makes pottery” in the English language, anyone who’s read the series knows that this isn’t the proper use for the word within the context. Instead of this out-of-place definition, Literāte offers up the proper usage based on context, which in this case, is the last name of the protagonist.
To learn more about Contextors, Literāte, and the future of natural language processing, we spoke with Steve Schuster, CMO of Contextors.
EdTech Times: Hello this is Hannah Nyren with EdTech Times and today I’m interviewing Steve Schuster with Contextors. So how is it going Steve.
Steve Schuster: It’s going great, thanks Hannah.
EdTech Times: So tell me a little bit about Contextors and what you do.
Steve Schuster: Contextors is a Tel Aviv based company. I’m located here in Boston and I”m doing business development for the company, but the main team, which is about 16 people is located in the Florentine neighborhood in Tel Aviv. What we’re doing is some natural language processing that’s resulting in an English language learning product that we’re calling Literāte. Basically, Literāte is learning enhanced reading. So if you think about reading just English text or content.
For somebody who is just learning English it’s very common for them to come upon a word they don’t understand or a context of the word they don’t understand and what we’ve done is using our technology is automatically extract all the information we know about the content and then re-embedded scaffolding for learning English back into the content. So it’s automatically extracted and embedded context specific definitions, grammar explanations, and then quizzes that are actually adaptive to what the learner is clicking on to learn about.
If the learner clicks on a word they don’t know the automatically generated quiz system remembers that word and will reinforce it later. Or, if they’re clicking on a lot of adverbs let’s say or words that end in ly they’re going to see that when they go to the quizzes they’re going to see a lot of that for reinforcing their acquisition of the language. So what happens is when they read along in a fashion called extensive reading they’re reinforcing the language over and over and to turn to what’s called intensive reading they don’t have to leave the context, they don’t have to turn to a dictionary or a book or to most importantly a teacher.
It’s really the world’s first digital product that replaces having a teacher sitting right beside you giving you all that information when you see a sentence “why do those words work that way, why does the English language work this particular way.” So it’s giving student a very, very effective way to learn English either in a classroom setting or on their own.
EdTech Times: Cool. So why was Contextors created?
Steve Schuster: It was specifically created to try and help people communicate.We understood that the world is very quickly turning towards English. There are about a billion and a half people studying English on some level in the world. If you do the math that is a really high percentage of the entire population of the planet. We feel if people can communicate better, there is a much better chance of getting along and resolving any differences between them. The founders really recognized that the existing technology for extracting the information in the text was really not accurate enough.
So today’s, what are called parsers are basically text analysis engines use a statistical analysis methodology that gets you to a 90% accuracy level. If you’re teaching somebody a language you’ve got to be perfectly accurate. I don’t want to tell you something wrong and have you go out on the street and speak wrong or communicate wrong. So, the founders created a rules based parser, which kind of was contradictory to what the prevailing science said was the right way to do it. But, being sort of innovative Israelis and maybe a little bit willing to go in a contradictory direction, said let’s try rules because rules are 100% accurate. So, having done that we applied it to this English language learning product, Literāte, so now we’re able to really have the accuracy and the very focused information that are two absolute ingredients for acquiring the language.
EdTech Times: Okay, great and one more question: What are you working on next in the future? How are you developing your product further in the next 12 months or so?
Steve Schuster: We’re scaling it up and scaling it out. Scaling it up means we’re adding a lot of content. We’re adding a lot more short stories because those are kind of bite sized pieces for us to process and for students to work with. So you get in there you read a short story, you click on words you don’t know, do the quizzes, and you’re done. You get the feedback and you can feel very good about what you’ve done. We’re also, in addition to scaling up with more content, we’re scaling out with more native languages. All the scaffolding I’ve described, the explanations of the grammar, all the quizzes and everything can either be in the learner’s native language, or in English. So, we’re going to scale out with more geographies and more native languages.
EdTech Times: Great. Well it was great speaking with you today, Steve.
Steve Schuster: Great speaking with you, Hannah.
A Texan by birth but a Bostonian at heart, Hannah is an educational writer, AmeriCorps alum, and one-time StartupWeekend EDU (SWEDU) winning team member. She started her career at a Pearson-incubated edtech startup, but has since covered travel, food & culture, and even stonemasonry in addition to education.