Google Analytics

Tuesday, July 29

Travails of a Lazy Kannada Learner

Old Kannada inscription dated 981 CE in Vindyagiri hill at Shravanabelagola.jpg
"Old Kannada inscription dated 981 CE
in Vindyagiri hill at Shravanabelagola" 
Based on my efforts to learn Kannada, I have come to realize that there is a disconnect between the resources available and resources required for learning the language. While there is a steady stream of apps to help Kannada learners, they all seem to be based on the same basic pattern which I believe is not useful.

When learning Kannada, a speaker of another Indian language is presented with two difficulties. One is the language and other is the script. In the absence of reading, the sole source of language data being absorbed is spoken. With spoken data, the learner can neither control the speed of the incoming data nor can he reliably repeat it. This severely slows down the vocabulary acquisition process.

Without knowing the script, he cannot read the newspaper which is a good cheap source of plenty of reading material. Reading the newspaper in a language also provides a good sense of the environment in which the language operates, allowing one to connect with the pulse of the city. This is a very important part of taking language beyond a mere transnational medium with handful of people.

But without knowing the language, learning to read is also difficult. One can always start with the approach of mugging up the character shapes, ligatures, diacritic marks, but how do you know if you read something correctly since you do not know the language. Also, in the beginning the process is so slow that it is easy to give up. One can try reading the boards on the road but at the vehicle speeds, it is next to impossible (might work in certain areas of Bangalore like Hosur Road!). Even today, by the time I am able to read "Sarkari" on the school boards, the boards are well past. It seems silly, but it is a real problem!

So how do we break this cycle? From my personal experience, I can suggest a two pronged approach that helps start a positive reinforcing cycle.

For the problem of learning the script, I made progress by focusing on the proper nouns. Since names do not change between languages, we can leverage our prior knowledge to make this work. Place names and boards on shops make an obvious candidate but as I have described above, shop boards are not suitable for a sustained study in the beginning. I believe person names are a better source.

So I turned to Wikipedia. While Kannada Wikipedia is much smaller when compared to English, you can still find articles that are available in both. Out of these, the most useful are pages that list names in a particular category. I started with Presidents of USA, following it up with Chief Ministers of Karnataka and so on. In 1 week, I made more progress than I had ever made with all the books promising me to teach Kannada in 30 days.

I have a theory about why this works. A list of names of Presidents of United States of America is at a sweat spot between what you know and what you don't know. Some of the names are very common, so you can figure it out by only reading the first 1-2 characters. With others, like a uncommon middle name, you have to work till the end. So there is a good balance of challenge and reward.

Additionally, here the characters are repeated based on the usage frequency. So the most common characters and their common variants are reinforced and little effort is spent on characters not so common. Of course the frequency is a little off with English names and so I switched over to Kannada names afterwards. I now buy a Kannada newspaper once in a while and try reading it. To my surprise, I find that a lot of vocabulary is pretty close to Hindi/Sanskrit!

On to the language part. Most of the apps, classes and other resources available focus on teaching the spoken Kannada. As far as I know that is also the recommended way to approach a language. I tried books like Learn Kannada in 30 days, Rapidex English to Kannada, even an old book written by Rev. F. Ziegler meant for British officers. But none of them worked. I also tried bunch of Kannada learning apps. They also proved more or less ineffective for long term sustained learning. Over time, their interfaces are getting more polished but the underlying content or ideas are not evolving.

There were some common problems. All these resources were based on collection of sentences, conversations for use in different circumstances. As is bound to happen with any made up conversation, they were awkward, forced and became useless very fast in real world situations. None of them provided any instructions or pointers for further reading, additional vocabulary etc. Some of them had Kannada text printed is such bad quality that it is useless for a beginner.

One book that has helped me tremendously though is called Conversational Kannada. Although this book is also focused on spoken language, there are some crucial differences. The best thing is that instead of dividing the material as per situations, it divides and arranges it in the order of increasing grammatical and sentence construction complexity. For example, in the first 10 chapters, you work with sentences without any verbs. Each chapter also has a theme for the conversations as well as introduces the informal phrases that are used in that situation. All the Kannada text is in roman, so you can read it without knowing the script.

But there was still one problem when reading the above book. Beyond the sentences in the book, I had no reading material available in roman Kannada. While I was able to read the script by now, my speed was not fast enough to allow me to read fluently which is required for vocabulary building. Also I needed to control based on my current vocabulary. For this, I turned to Twitter.

With the ability to see the conversations, Twitter is a goldmine for a language learner. Plenty of people use roman Kannada on twitter. You can search for specific words and find conversations where that word is being used. Best part is that the conversations are natural, there is plenty of English mixed in allowing you to follow the conversations and more reading material is being created all the time. People are helpful, so you can even ask some of them what a word meant. I did this for many days. There are challenges around search - some words are common in other languages as well, some are spelled in multiple ways, some others may not have been used on Twitter in recent past. But overall, it was very helpful.

So combined with the above two, I am confident that I will be able to read some simple books in Kannada by the end of the year. Both the ideas that helped me can easily be packaged as apps and provided to the new learners. I did all the work manually - finding parallel Wikipedia pages in English and Kannada, searching the Twitter conversations. All this can be handled by a centralized service that can collect & process the data and then serve it to the learners based on their requirements. Based on my experience, it will prove to be a lot more effective than the cookie cutter apps that are available now.

Since then I have discovered one more resource that is Quilpad Switch. This tool allows you to view any webpage in the script of your choice. So now you can read any Kannada page in roman or in Devanagari script thus immediate access to tons of reading material. This may even reduce the need of the Wikipedia pages that I used. The requirement is now for a curator that can point out the texts based on the current vocabulary of the learner. Another idea ripe for making into an app. :)

Image by en:User:Dineshkannambadi Original uploader was Dineshkannambadi at en.wikipedia - Transferred from en.wikipedia; transferred to Commons by User:Papa November using CommonsHelper.(Original text : Uploaded by creator). Licensed under Creative Commons Attribution-Share Alike 3.0 via Wikimedia Commons - https://commons.wikimedia.org/wiki/File:Old_Kannada_inscription_dated_981_CE_in_Vindyagiri_hill_at_Shravanabelagola.jpg#mediaviewer/File:Old_Kannada_inscription_dated_981_CE_in_Vindyagiri_hill_at_Shravanabelagola.jpg