Data-based language learning
Optimising the language learning process with statistics, data and logic.
There’s a lot of cliches in language learning.
It is fairly well known that you can’t learn a language in X amount of time, and that there is no best way of learning languages for everyone, but what if we could find a way to improve our learning outside of these two paradigms?
What if we could learn languages in the most efficient way - making the most of our time allocated to language learning?
My hypothesis is that whilst there is no singular way to learn languages, there are ways to make the process as efficient as possible. In this blog, I will be exploring this hypothesis and providing tools, concepts and ideas to help language learners structure their learning style and path, as well as provide useful case studies of the application of this idea.
The ideas for this framework have come to me over a decade of self-taught language learning. I began my journey when I was 17, and got a scholarship to learn Mandarin in China for a year, before doing a degree in Chinese and working in China for a few years. Having access to China’s international community, I saw an opportunity to learn a lot of other languages and practice them, and it’s an opportunity I fully seized. In that time I picked up some Russian, Korean, Japanese, Shanghainese, as well as a smattering of others such as Swahili, Arabic, Hindi and Farsi.
As I began learning and using these languages, certain patterns emerged, and after some deep thinking of a way to structure this intuition, it came to me in a dream that there’s a way of optimising the process of learning both one language, and multiple languages by looking at languages statistically.
6 years ago, I decided to return home to Wales to try and build something out of this idea. I did an MBA in International Business to give me as good a shot as possible, and I set up Pai Technologies with my Co-founder, Simon. We have spent the last 4 years trying to build company around language data, and a language learning platform based on these ideas. Covid and other events over the last few years have made this a difficult and slow process, but the idea is too exciting to let go of.
The goal of this blog is to explore the possibility that there is a way of optimising this process, and in doing so, save people a lot of struggle, difficulty and time when it comes to learning languages - one of the most rewarding things someone can do.
We will explore this possibility by applying data, statistics and a bit of logic to language learning, especially self-taught language learning.
My current stance is based on personal experience self-teaching languages. I have had no formal education in language education, but I have read many of the key works and understand some key concepts. I am happy to hear criticism, but I hope critics can see that I am doing this from a good place and in the spirit of open inquiry.
Below are my current presuppositions, as far as I can recount them, that have given me the idea that language learning can be optimised:
Language learning is a process, and like all processes, there are ways it can be optimised.
Language can be approached like a skill, and like all skills, unless it is essentially innate (like one’s mother tongue), it’s better to learn some concepts before you can use it properly. Like learning to ride a bicycle, it’s quicker to know what the pedals do, and how they work, before you get on. You can learn by trial and error, but it’s going to be painful.
A skill-based learning model provides that you move from declarative (you can say the knowledge as a fact), to procedural (you know how it works), to autonomous knowledge (you can use it naturally).
Knowing a language requires knowing the phonetics, vocabulary, set phrases and structures, grammar and on a higher level, culture of said language.
Having autonomous knowledge of the vast majority of a language you will come across is true ‘fluency’.
Fundamentally, you teach yourself languages. You can’t rely on others to do the leg work for you, even if they are perfect teachers.
By optimising the process of language learning, we can ensure that the time we spend actively learning a language can be used in the best way for maximum results.
By looking at the interplay between what the learner knows, what they need to know, and how they learn what they need to know, we can find an optimal path through vocabulary, phonetics and grammar for each learner.
Complex parts of learning specific languages can be broken down to provide a “minimum effective dose” of active learning.
Languages are not inherently hard or easy - difficulty is a matter of how different a language is in terms of phonetics, vocabulary and grammar from the languages a learner already speaks.
I will be exploring each of these points in more detail going forward. I look forward to going on this journey with any readers I should pick up on the way. Please consider subscribing below.
If you would like the opportunity to support my company in bringing this style of language education to potentially millions around the globe, as well as help create materials for some under-served languages in the process, get in touch through the company website - We are looking for investors for our first round and are looking for people who share our vision.
If you want to talk to me about any of these ideas, please join the discord server - I’m obviously very busy with work but I’ll try to reply to anything sent and always happy to discuss new ideas.
Diolch i bawb am eich cefnogaeth.
Thank - to - everyone (soft mutation following ‘i’) - for - your (pl) - support