A computer scientist is studying ways to express emotions in automated translations of Urdu.

You may not know much about Urdu, but it is spoken by around 65 million people, mainly in India and Pakistan. It is quite similar to Hindi, and the two tongues taken together represent one of the world’s major languages: according to some estimates, the Hindi-Urdu language group is second only to Mandarin Chinese.

As you can imagine, however, Urdu differs vastly from English – and thus confounds machine translators even more than European language pairs.

For this reason, a researcher is working on how to improve automated translation from Urdu by indicating the emotions inherent in the language.

In an NPR report, computer scientist Rohini Srihari says that computer translation systems for Urdu are “often too literal.”

At the University of Buffalo, she is thus developing a computer program that she claims can identify the emotional nuances of written Urdu.

“On the screen, you can mouse over a section of script, and if it carries a ‘negative’ connotation it will highlight red,” explains NPR journalist Christopher Joyce in the article entitled Computer Translator Reads between the Tweets. “If it has a ‘positive’ sentiment, it glows green.”

That all sounds a bit simplistic, however. Indeed, the article also presents arguments against the concept of computer-programmed interpretation of emotions.

Betty Carlson

See Also

• Language Translation, Inc.
  A leading language translation and interpretation service provider for more than 20 years.