1. Forum
  2. >
  3. Topic: Duolingo
  4. >
  5. Intralingual (English to Engl…


Intralingual (English to English, etc.)

It's probably been asked for many times. But I want nevertheless to suggest, especially in the age of the incubator, that Duolingo open one more door to learning:

The Intralingual.

Here is one argument I offered to a native Urdu speaker I know who has devoted more than 30 years of his life to teaching Urdu to kids (and now kids of kids) of rich Urdu-language heritage who are growing up in Urdu-poor Texas:

Adding Urdu to Duolingo would greatly benefit hundreds of thousands of people like myself who have a tolerable fluency in spoken Urdu, but lack functional reading or writing skills. Of course the number of people is really in the millions -- but most of them do not have access to a computer or a smartphone.

Perhaps some day it will be easy to teach Urdu literacy to all the illiterate children of Pakistan, simply by giving them access to a Duolingo app on a portable device.

I humbly put it to you that there are children in every country, especially in America, who could improve their speech and writing in their native tongues with Duolingo technology.

The trees/paths might look different for intralingual learners. But I believe those trees would bear wonderful fruit.

October 11, 2013



How do you imagine improving native language skills with Duolingo technology? Duolingo method is based on translation.


In terms of literacy it could probably be done similarly to the way duolingo will teach complex logographic languages like Chinese and Japanese. Although teaching someone to read for the first time is a lot more difficult than teaching them a new script.

It doesn't seem to be really part of duolingos business plan or remit to be such a foundational learning tool for people but it'd be a worthy thing to take on at least.


100% Literacy rate is brought about by sharing the skill with people who don't have it. My brother taught me how to read, before any school. My friend from India taught me how to read Devanagari script, it was quite easy to learn with the support, just having someone with which to ask questions. You shouldn't try and leave such an important job in the hands of a computer , that is a pretty bad dereliction of duty. Rather, it should be a cultural norm for every single literate human being to share that skill with at least one other person in their lifetime. If we all took it upon ourselves to do that, illiteracy would disappear overnight, but instead we write off those who can't read as uneducated, backward and 'not our problem'. Would anyone let a man starve on their doorstep while they were inside eating a delicious soup?


It's an interesting idea, but I really don't understand how you would implement it. Any ideas?


I've actually spent some time thinking about this because it's an interesting idea but would be difficult to implement, but I think the following drills might be able to be enough to build a course:

Picture-definitions word definitions (select the right definition of a word) correct the grammatical mistake correct spelling mistakes audio questions (same as for current courses)


Thanks, Olimo and Ontalor, for your comments. I think there are several Duolingo tools that would work well "intralingually."

Consider that an intralingual learner may need to improve reading skills. May need vocabulary building and spelling improvement. May need improvements in syntax/grammar.

Duolingo could build vocabulary with picture drills, just like the existing drills but intralingual. Those drills would help build reading skills, but so would drills where Duolingo speaks a sentence, and the listener chooses word-tiles (an app drill currently that could be used on-line, too) or the listener types the sentence (app and on-line). All of these would build spelling skills, too. The latter would introduce grammar concepts and reinforce them.

Duolingo has drills now that require selecting all the right translations. Intralingual could do that for spoken sentences...

...but the real coup would be intralingual questions that ask for the student to answer questions. Those answers would have to be grammatically sound, etc.

That type of drill would also be great in duolingo, if it does not already exist. Ask, "Comment çava?" And the student has to choose correct replies or assemble one from tiles or write one from scratch. We can easily imagine it in duolingo now, but it would work intralingually, too.

And as for the bottom line, Duolingo's incubator is the perfect cost-saver. All volunteer driven. Every exercise created intralingually could be mined for duolingo purposing. And every child or adult whose horizons are expanded by intralingual becomes 1) a potential duolingo user, 2) an ambassador for all things duolingo, 3) a person who contributes more to the world than before, 4) perhaps all of the above.

And Luis is a smart guy. He can figure a way to achieve Re-captcha revenue streams in both intralingual and duolingo. It's just a new drill away, I'm sure.


I don't dislike your idea by the way, I just don't think it is as trivial to implement as you think it is. Duolingo's team is already burning the candle at both ends just fulfilling the promises it has already made. If you can find a way to use the incubator to achieve what you are thinking of, more power to you, but I think that the challenge that you are trying to solve is different, and needs more than what Duolingo can offer. What your idea says to me is that it would need a chat-bot that would be able to evaluate an infinite variety of spontaneous responses as to whether they are grammatically correct or not.

To keep the subject in line with 'comment ca va', a pretty basic interrogation in any language , well you can answer the question 'How are you' with anything from 'Fine, you?' to 'Well, considering that a badger is chewing on my toenails, I would have to say that I am feeling mildly uncomfortable'. I grossly exaggerate the point, but you can see how hard it is to build a computer program that can fairly predict every possible response, without missing some unexpected but perfectly valid ones. Every single response has to be explicitly entered and explained and rated before you can call it a fair test, and while this is evidently possible for translating isolated canned phrases directly into another language without shifting their meaning, is it even realistically possible to extrapolate that method to spontaneous, random conversation which can lead in almost any direction?

I'm sorry by the way, I really hate so being negative, and I would be quite happy if you said something that made my doubts seem unfounded :)

Learn a language in just 5 minutes a day. For free.