Researchers at Stanford University have developed a computer simulation that mimics the brain processes used by babies when learning about speech. 

The program can decode sounds from different languages in much the same way as a baby, offering new insights into how we learn to understand speech.

"The debate in language acquisition is around the question of how much specific information about language is hard-wired into the brain of the infant, and how much of the knowledge that infants acquire about language is something that can be explained by relatively general purpose learning systems," said James McClelland, a psychology professor at Stanford.

Professor McClelland claimed that the computer program debunks existing theories suggesting that humans are born with a hardwired knowledge of all sounds and languages.

He also believes that it supports the idea that babies continually sort through the sounds they hear to develop an understanding of the structure of the language.

"People have tried to argue that it was not possible for any machine to learn these things, and so it had to be hard-wired in humans. Those arguments, in my view, were not particularly well grounded," said Professor McClelland.

"The problem the child confronts is how many categories there are and how it should think about it. We are trying to propose a method that solves that problem."

The team provided the program with recorded 'conversations' between mothers and babies in English and Japanese. The software analysed and learned basic vowel sounds at the same rate as the infant human counterparts.

Professor McClelland and his team have concluded that if a computer program can be taught this way, then so can a baby.

Details of the research appear in the Proceedings of the National Academy of Sciences.