Children from poor homes should learn to read using techniques developed to tackle dyslexia, rather than focusing exclusively on current programmes of synthetic phonics, according to a University of Cambridge professor.
Usha Goswami, director of the Centre for Neuroscience in Education, said that existing phonics programmes failed to take into account the difficulty that some children had distinguishing individual sounds.
Last week, she was awarded pound;365,000 to explore whether poor children could improve their reading skills more rapidly by using a computer game that tests awareness of longer sounds within words, which was originally developed to help dyslexic pupils.
The money from the Education Endowment Foundation (EEF) and the Wellcome Trust will fund a randomised controlled trial with 400 six- and seven-year-olds, to see if playing the game for 10 minutes a day significantly improves their reading skills.
Current phonics programmes encourage children to identify phonemes, the smallest units of sound, which are put together to make up syllables. But a level of sound between phonemes and syllables - called onset and rime - is missed out. This is what the game encourages children to focus on.
"A syllable is a package of sounds," Professor Goswami said. "If you break the word into the smallest unit of sounds, cat becomes c-a-t. But [using onset and rime] it becomes c-at. Synthetic phonics, if you do it exactly by the book, doesn't teach this level."
She added: "The computer game was developed for dyslexic children but [its use] suggests it should also be helpful for disadvantaged children generally, who typically have impaired language and reading skills. And indeed, the game should be helpful for all children in terms of teaching English phonics."
The government has pushed the use of synthetic phonics in primary schools by introducing a phonics check at age 6 and funding resources and training. Ofsted has also sharpened its focus on phonics in routine inspections.
The use of onset and rime in teaching children to read has been controversial, with advocates of synthetic phonics saying onset and rime adds an unnecessary layer of complication. But Professor Goswami said one of its biggest advantages was that spelling was more consistent when using longer sounds.
The computer game, GraphoGame Rime, which was developed by a university in Finland, encourages children to match sounds they hear through headphones with groups of letters they see on a screen. It analyses their answers and provides feedback to teachers.
Greg Brooks, emeritus professor of education at Sheffield University, said debate was heated over whether learning to read was made easier by blending large units of sound.
"This latest grant from the EEF will fund what is probably going to be the most rigorous trial yet of large-unit phonics.so it will be very interesting to see in a couple of years what the results of this trial show."
Dr John Rack, director of education and research at Dyslexia Action, said: "From our point of view, we've always recognised that some kids don't.hear all the small sound segments clearly, so for them it is better sometimes to use bigger chunks.
"We agree it may not be the best way to teach all children - using phonemes may be better. But for those who don't get it we need to be more flexible, and units which are more consistent and easier to distinguish do play a part. We have an intervention programme, and onset and rime work is in that."
Janis Burdin, headteacher of Moss Side Primary School in Leyland, Lancashire, says using different phonics techniques has benefited her pupils.
"We don't teach Letters and Sounds [a synthetic phonics programme] exactly as it should be taught," she explains. "And we do use onset and rime with word families like cat, sat and mat.
"We have got away with it because we have such good results. I would argue that the way we do it, our children respond better."
In 2013, Ofsted praised Moss Side Primary as outstanding and said pupils' achievement in reading was particularly strong. Eighteen per cent of its pupils claim free school meals, yet 89 per cent reached the expected level 4, compared with 86 per cent nationally.