Decode Chinese: We are building the future of language education (Kickstarter coming soon)

There are a few good Taiwanese series on Viki. Depending on the kind of stuff you are into.

Viki has very few content with Taiwanese IP. If you’re in Europe it’s great. Even better with an old 9.99 USD per year basic membership.

YouTube and iqiyi have caught up a lot with usable subtitles the last 2 years. Before Viki was the only source of Chinese Drama with good subs (as Netflix had next to nothing)

Hello Ironlady and thank you for your reply. I am sorry to hear that you feel mistreated, that was not my intention. I thought that Jinyu response was thorough and captured most of the feedback and concerns in this thread (which actually builds on almost everything you have written in your previous posts).

My team has learned a lot of from the feedback and we realize that our presented message is severely misunderstood. The primary issue is the misconception that Decode is just a “Lookup tool” + “Flashcards”. I can see why people might want to think that based on our presented materials because we attempted to try to make it easy for all people to relate. It is important to understand that a Kickstarter campaign usually requires short videos and description because most people usually lose interest after 30 - 60 seconds. Therefore I decided to only attempt to provide a brief overview of everything (like using big titles and short description because most people just quickly scroll all the way down to the end of the page). This is difficult and I am not suitable to do this myself, thus why we will onboard a marketing and design specialist to help us out.

All I can say before we publish our new presentation is that in summary:

Decode is a platform we have built that uses our own resource description framework to manage structured linguistic resources on a knowledge graph. This enables everyone to build new services that uses the same standard to ensure all data points through the students learning journey can be efficiently combined and utilized. This includes all four domains such as speaking, reading, writing, and listening but also cultural context. Decode is not just Decode Atlas and Decode Study, those are just two of the services we have started to experiment with on top of our platform.

Decode Atlas is not a lookup tool like a regular dictionary. Decode Atlas is a user-friendly interface for both students and teachers to view, manage, and build learning resources on our framework. This includes mediums such as video, audio, and text and resources such as words (lemmas), grammar, sentences, and cultural context/dialogs. If you look at our demo video/website at the Decode Study section, you can see that the “user” only selects to build a Word Card. However, if looking closely, there is a bunch of other resources from Decode Atlas in that list. Once again, we are engineers and we are not particularly good at marketing.

Decode Study is not a traditional flash card system like Anki or Quizlet. However, this is ultimately my fault because I decided to call it a supercharged flash card system. The reason for this that I thought this would make it easier to understand for the general population. What Decode Study actually does is to build a personalized learning journey by stitching together all your learning materials (school, work, and everyday life) with help from structured linguistic data and resources through Decode Atlas.

Because Decode Atlas breaks down and connects languages on the smallest level possible, we can track exactly all data points necessary to ensure a smooth learning experience. You can kinda see Decode Study as your study companion and Decode Atlas as Decode Studys knowledge database that Decode Study can utilize to map out and prioritize exactly what you need to learn. Decode Study is also more of a learning game than a flash card learning tool. Not only can Decode Study be used to practice word and grammar lists from your course books but also mix in cultural context and dialog to practice real scenarios.

As I mentioned earlier, our initial goal was to build a personalized AI powered language teacher (although I prefer data-driven and statistical models) by emulating the real-life experience. But due to the lack of structured data and resources, we were first required to invent the framework. With our platform you can even do things like select where you want to geographically learn Taiwanese Mandarin (Like in Tainan) and Decode Study will help you to understand the local environment and customs (See the video covering a cultural context card on the Lunar Festival in Tainan). But also help to prioritize the learning materials based on your goals and interests.

I think one of our mistake was being too technical and not simplify our solution by for instance anthropomorphize Decode Study into an avatar. Our design today has obviously an engineered feeling with only bare-bone interface which will soon change to a more user centric design with simplified concepts. All I want to say is that calling our solution just a flashcard + lookup tool is severely diminishing our research spanning more than two years. But once again, not your fault to think that based on the information we have presented.

Please understand that how much both you and other people in this thread has helped us to improve and prioritize our marketing in the future. If you want, I would love to also talk with you Ironlady and show more when the time is right.

Based on my previous posts, I am not sure if it matters to explain everything through text format alone. Hopefully we can show the new presentation instead soon.

Until we announce our updated presentation through our newsletter (you can stay updated by subscribe through this link) I just wanted to at least attempt to clarify the general misconception of our solution.

// Alexander, Decode

1 Like

If you live in Taiwan and are learning Mandarin or Taiwanese, then eavesdropped conversations are a form of comprehensible input too. Just don’t go around pressing strangers in the belly button expecting them to repeat the dialogue :slight_smile:

Thank you for the article and the description of the “cold character reading” approach. I had been wondering whether students could effectively learn to read Chinese characters by going straight into Chinese texts the content of which they are familiar with when presented as audio and/or pinyin. without having previously learnt to write by hand a single character. From your experiments it seems they can do this without first having had even any explicit explanation of radicals.

In my own case learning Chinese many years ago, I acquired decent proficiency in the spoken language (using pinyin as a reference) before attempting to read characters. I could also write by hand 2000 plus characters, associating each charter with a particular word (using the Heisig mnemonic method I was able to do this in about a month). With this background I was able to quite rapidly start reading native level content written with characters. The process was relatively easy because context allowed me to fill in the gaps of unknown characters. The phonetic components of Chinese characters also helped greatly (but not, as many learning materials presume, the meaning radical, which is almost always too nebulous to be of much practical assistance). For instance: learning to read Chinese text with characters if I came across a sentence with an expression I would recognize as audio/pinyin, e.g. “suffer a set back” shou4 cuo4zhe2 I would probably -triggered by the phonetic radical, read the unfamiliar character cuo4 as zuo4, but would be able to immediately correct to cuo4 because that is the only possibility in that context. This combination of context and hints from the phonetic component of characters allowed for rapid mastery of reading Chinese characters.

I was wondering about was whether looking back I might have skipped that month of learning to write characters by hand (which itself might take years if I’d used traditional route methods). I absolutely agree with you that being able to write large numbers of characters by hand is little more than a parlor trick. With smartphones, you can always look up characters on the few occasions you might want to laboriously write something by hand: some form, or a love letter or a a get well card.

I also wondered whether the phonetic components of characters might be studied systematically: you know, learn characters with the same phonetic radical and identical or similar sounds. Again, my hunch is, and this seems to be corroborated by your teaching experience and research, is that it would be better to allow students to make use of this phonetic triggers instinctively, taking advantage of the context of full sentences and text, rather than learn them as word lists.

I’m guessing that many or most Mandarin learners would cite learning to read Chinese characters as their biggest challenge (second only to learning to write characters by hand - which I suggest they shouldn’t), so if Cold Character Reading or a variation upon it can help students acquire proficiency in reading characters much faster than traditional methods then this will be a great boon for Chinese teaching and learning.

1 Like

I think we do understand that Decode is much more than a simple flash card system. It allows you to track and link material you’ve previously learnt, for instance the “hand” radical in “wo3” with other characters you’ve learnt with the same radical. It contains all kinds of references, tailored to the individual, about local usage of words, culture, etc. It seems to provide all kinds of advanced AI feedback and linkages between previously learnt material. And so on.

It is a lot more than a flashcard system or a lookup dictionary, but, in my opinion, and this is the crux of the matter, it seems to be a lot more in the wrong or rather in a peripheral direction. The crucial factor determining proficiency is exposure to comprehensible input - massive amounts of audio (whether in an app or face to face) and reading (of complete dialogues, stories etc rather than of isolated words or isolated sentences). And I don’t see where Decode (the name itself implies a misapprehension about the way languages are learnt) contains large quantities of original audio and text (as opposed to example sentences). Unless Decode is claiming that it provides an approach that bypasses the need for some kind of comprehensible input as the main element of language learning (a claim to be dealt with separately, if it is in fact being made) then the student is still going to have to look elsewhere for comprehensible input materials. They might be able be able to put dialogues from their textbooks etc into Decode system and access further notes and explanation and link words in the dialogues with previously learnt words etc etc but this, again, means that Decode would at most be a useful supplement on the side, and in practice would be an unnecessary distraction to beginner and intermediate students.


The key question is: what “real-life experience” is the product seeking to emulate, and if it’s emulating a current real-life experience, how is it a breakthrough new concept in language learning (acquisition)?

I don’t think your mistake was being too technical, with this audience at Forumosa. I think the mistake was not providing actual technical information on the language/linguistic side. This isn’t a general audience – the Forumosa audience is people who have been engaged in Chinese learning and/or teaching for a long time, and who have qualifications, can read and evaluate research, and don’t want to read the marketing hype, but rather to know what the real function and theory is.

I’m sure everyone here will be interested to learn more when your group feels that it’s able to provide that kind of information. Critical or not, we’re interested. Everyone would love to have something that actually fulfills the marketing text, but we need to see more than that marketing text to buy into it.


Evesdropping on my coworkers conversations usually causes me to regret learning Chinese in the first place.

1 Like

Also being told i was 90% baiting when i asked for some money where the mouth was on research really rubbed me the wrong way. I’m mostly checking in now to see your posts on this thread.

Thank you for sharing this! It backs up what my first Chinese teacher did back in 2009 and explains why my further Chinese study was such a struggle (those teachers insisting on us learning all parts of Chinese, all at once). I will be referencing it as I move more towards probably teaching Chinese in the US. I am really interested in making Chinese language accessible and so tired of people insisting that it’s “too hard to learn”, especially when it’s one of the only languages you’ll encounter that becomes really just a matter of “learning more words” once you’ve got a foundation.

1 Like