[an error occurred while processing this directive]
BBC News
watch One-Minute World News
Last Updated: Wednesday, 17 May 2006, 23:55 GMT 00:55 UK
'Big brother' informs baby talk
11 cameras record any waking activity in the house

Every movement, gurgle and chuckle made by a baby in the first three years of its life is being recorded by a scientist in the US.

Professor Deb Roy of Massachusetts Institute of Technology (MIT) is recording his son's development to shed light on how babies acquire language.

The Human Speechome Project, as it is known, uses cameras and microphones installed in the scientist's home.

The project will eventually gather 400,000 hours of material.

"As every proud parent knows, there's no such thing as too many images and videos you can take of your newborn," Professor Roy told a press conference.

"I think we're taking this to a whole new level."

Language difficulty

There is still a considerable amount of debate about how infants acquire language.

Although listening to the cooing of parents is thought to play an important part, most scientists believe it cannot be solely responsible for the rapid progress seen in most children.

Professor Roy (copyright Webb Chappell)
There are numerous spin-off opportunities beyond the Speechome
Professor Deb Roy
Instead, language-specific genes and environmental factors have both been put forward as additional factors that help children to learn to speak.

Until now, the environmental influences on development have been very difficult to test because scientists have been unable to observe a baby for long enough in its home environment.

The Speechome Project will change that by generating and analysing vast tracts of recorded material.

For example, to understand how Professor Roy's son learnt his first words, the scientists will be able to mine their records to see who used those words around the child, where they were and what the child was doing at the time.

Frank Moss, director of MIT's Media Lab, believes the project has close parallels to the Human Genome Project.

"Just as the Human Genome Project illuminates the innate genetic code that shapes us, the Speechome Project is an important first step toward creating a map of how the environment shapes human development and learning," he said.

Big brother

The project started recording nine months ago when Professor Roy's newborn son left hospital.

Since then a "big brother" network of 14 microphones and 11 omni-directional cameras has been recording his son's waking hours.

Infant holds a parent's hand
Interaction with parents is thought to affect language development
The surveillance system is turned on at eight o'clock in the morning until 10 o'clock at night, producing nearly 350GB of compressed data every day.

It will be switched on for the next three years, by which time Professor Roy's son should be using complex language and spending more time outside, making recordings more difficult.

In case Professor Roy's family requires some privacy, every room is fitted with a PDA that can turn the microphones or cameras off.

An "oops" button allows people to erase the last few minutes of footage.

"You can type in how many minutes back in time you want to scrub permanently from the house's memory," said Professor Roy.

Personal video

After the data has been collected, it is temporarily stored at the house before being sent to a massive petabyte (one million gigabyte) disk storage system at the Media Lab at MIT.

There, both humans and computers are crunching the data to look for patterns.

However, Professor Roy is keen to stress that most images will never be seen by human eyes.

Mother talks to her infant
The project could produce the ultimate family album
Instead, software will process the "immense flow of data" so that common actions such as doing the dishes or changing a nappy can easily be recognised by the researchers.

Other tools analyse speech patterns or show how people move through the different rooms in the house.

Together, the different systems will build a complete picture of all the stimuli that the infant experiences, allowing the model to "step into the shoes" of Professor Roy's son.

The team then hopes to build computers that can learn words and grammar, from hearing and seeing precisely the same images and sounds as the child, to understand the learning process in humans.

As well as these insights into language development, Professor Roy and his team believe the technology that has been developed for the project may also have applications in other fields such as personal video or analysing images from security cameras.

"There are numerous spin-off opportunities beyond the Speechome," he said.

At least one of these is the ultimate family album for his son when he grows up. But Professor Roy says that sitting through hours of baby photographs won't be a laugh a minute.

"Most of the recordings are pretty boring."

Children create new sign language
16 Sep 04 |  Science/Nature
First language gene discovered
14 Aug 02 |  Science/Nature


The BBC is not responsible for the content of external internet sites


Americas Africa Europe Middle East South Asia Asia Pacific