Non-khmer speaking netizens around the world may be able to read and understand Khmer-language websites as soon as the end of the year, if the planned release of Google’s Khmer translation feature goes according to schedule.

The Khmer translation tool has been in the works for about five years now, said Heng Sovann from the Global Developer Group, an IT software developer involved in the project.

Sovann told the Post an alpha version was already available and is now being tested internally.

“We are still trying to solve some quality problems, like accuracy problems, which affect the translation quality,” he said.

There are problems because the way Google’s translations work is by “brute force”, explains Javier Sola, program director at Open Institute, an information and communication technologies NGO here that has been working with Google for the past year.

“It is purely statistical. You take millions of sentences written in English and Khmer, and put it in a sort of washing machine,” Sola said. “This forms a database that uses these sentences as examples when they have to make a translation.”

The bigger the database, the more accurate translations will be, he added.

But the existing repository of English and Khmer translations is not as comprehensive as say, English to French translations.

“This will take time to build up,” said Sovann, who says the accuracy of translations will increase with time.

Experts say this feature would contribute to greater understanding both in and outside Cambodia.

“People that don’t speak Khmer would be able to use this new feature to understand messages from Cambodians,” said Tony Lim, the US Embassy’s information management specialist. “It’ll be one way to get the word out.”

Lim spoke to the Post on the sidelines of Tech Camp – a technology conference for NGO representatives organised yesterday by the US Embassy.

Google did not reply to queries by press time.