Letter use analysed to date ancient Irish texts

Fri, Jan 13, 2012, 00:00

A DUBLIN-BASED student has taken the Irish language, mashed it through a wringer and in the process has come up with a powerful tool for analysing ancient Irish texts. In particular it looks promising for dating texts of unknown but ancient origin.

Aoife Gregg (16) is presenting her findings at the BT Young Scientists and Technology Exhibition under way at the RDS. Although the exhibition is open tomorrow, it reaches its climax this evening when the young scientist for 2012 is announced.

The transition-year student from Loreto College, St Stephen’s Green, carried out a letter frequency analysis of the Irish language, using a computer to determine which letters occurred most frequently. The top four in order were A, I, H and N, but then she got the idea to measure letter frequencies in older documents to see how the Irish language changed.

She studied 17 documents dating as far back as AD 600 and immediately found changes in the letter frequencies. The letter A is much more frequent today, she said yesterday and also the letter H. This may seem a bit academic but in fact she used these methods to date old Irish texts. She correctly determined the ages of eight out of 10 old texts, she said. “It provides an alternative way of dating Irish documents.”

Eamon Hennelly McCarthy (16), and also a transition-year student, conducted a different analysis using non-invasive methods to identify people. A student at Skerries Community College, Co Dublin, he used “soft” biometric details such as hair colour and type, tongue folding, gender and other cues to correctly identify individuals.

While fingerprints are a definitive identifier, it is quite possible to use a wider selection of soft biometrics to do the same job, he said. He developed a check list and computerised the analysis so identification came down to getting a match between stored biometrics and the individual. He settled on a list of 15 attributes, he said. “The more biometrics I added the more people I could identify. I got a 96 per cent accuracy.” Of the four he did not identify, two were twins and the remaining two could have been identified if age had been included as a biometric. He left out age and similar attributes that changed radically over time.

A project by Robert Gabriel (18), a sixth-year student from Scoil Mhuire Gan Smál in Blarney, Co Cork, also depended on computers, but this time to make them run faster. He has developed a piece of software, an algorithm, to help make the internet go faster. “It helps your download speed in areas where you can’t get good connections,” he explained.

The algorithm smoothes out intermittent connections by breaking up the downloaded information in a different way, making it possible to take jumps and pauses out of the picture. And in areas where connections run well, it speeds them up still further, allowing information to be displayed almost instantly. In tests his system ran faster than search engines Bing and Yahoo, he said.