Some months ago, Lalitesh Katragadda, head of products at Google India, was capturing images of small retail outlets at his grandmother's village near Vishakapatnam in Andhra Pradesh. An old woman who was selling red gooseberries on a pushcart asked him what he was doing. He said he was taking pictures of shops to put on the internet for others to see. She instantly said, "Why can't my business, my push cart too?" For Katragadda, this was further reinforcement of his belief that people were hungry for recognition and identity. He saw it in cities, amongst the educated, but equally among the uneducated in villages.
"That's why a large number of people of very unassuming profiles are taking to You-Tube , irrespective of literacy or language barriers," he says. "Images and videos evoke powerful emotions for the educated and uneducated alike. You will be surprised to see the profiles of new-comers to YouTube --farmers , drivers, mechanics, village housewives, and people of simple trade who are semiliterate or illiterate."
Katragadda is now convinced that video is the way ahead for the internet in India. In the past few years, internet companies, including Google , have had a big focus on creating and enabling local language textual content with the idea that such content would deepen the penetration of the internet in India. They developed transliteration technologies that allowed people to use the English script to write the local language words phonetically and the software would automatically convert it into the local language script.
They developed translation technologies that enabled the translation of an entire document. Wikimedia Foundation, which operates Wikipedia, has spent a fair amount of energy trying to get people in different states of India to try and translate its English content into local language content, as also create fresh local language content.
But these have so far had very limited impact. Transliteration is useful only for specific people in specific circumstances . Machine translation invariably came with mistakes, requiring significant manual intervention. And Wikimedia Foundation has been struggling to find volunteers to create sufficient volumes of local language content.
So the internet in India remains restricted largely to the English-speaking middle class. The user number is estimated to have touched 121 million last year, about 10% of India's population. Taking the number significantly beyond this requires a big move into small towns and villages. Here, local language dominates, and raising internet numbers through textual content looks an uphill task.
Vikram Dendi, director of product strategy & marketing in the machine translation group atMicrosoft Research, Redmond, says technology has to become more and more natural , has to get closer to what people are comfortable with. The company is looking at more visual and tactile ways in which to reach out. It's still putting an intense amount of effort into perfecting machine translation, and that exercise is now proving to be useful beyond simply creating local language content (from English content).
With its translation app on the Windows phone, you can point the phone's camera at any text and have it instantly translated to the language you desire (the accuracy is high when the content for translation is short, like signboards or restaurant menu cards). The translation technology also helps in converting speech from one language to another.
Again accuracy is high when the sentences are short. All you need to do is speak into the phone, and the application converts it into speech in another language. "My wife is German and her mother does not speak English. So I sometimes use this application to speak to her. It's not entirely accurate, but it's useful," Dendi says.
Yahoo says it has a big focus on bringing entertainment --primarily cricket and celebrity -- video content in local languages.
So video and more natural interfaces, rather than plain text, may be the answer to creating India's next hundreds of millions of internet users. Google's Katragadda notes that many middle class children today grow up learning nursery rhymes, animal sounds and the alphabet on YouTube rather than from books. Online video is their primary interface.
However, video consumes a lot more bandwidth than does text. And India's low broadband penetration is currently a huge limitation. But the government's ambitious project to give broadband connectivity to 250,000 villages and cheaper 3G rates could significantly lower the barrier in times to come.
No comments:
Post a Comment