New Delhi: Two Indian scientists have discovered a common hidden pattern lurking in the use of sounds in multiple languages that they say could serve as a novel mathematical tool to study undeciphered scripts.

The researchers at the Institute of Mathematical Sciences in Chennai have found that words in diverse languages from across the world start with a far greater diversity of sounds, represented by letters or symbols, but end with relatively restricted sets of sounds.

This "novel" pattern of how people begin and end words appears shared by over 20 present-day languages, including Hindi, Arabic, Chinese, English, Hebrew, Malay, Urdu, Russian, Turkish, among others, and three ancient scripts - the Egyptian hieroglyphs, Sumerian cuneiform and ancient Greek.

The scientists, who examined the positions of the 26 letters of the alphabet in about 97,000 unique English language words, have found that almost all letters - X, Y, and Z being exceptions - have a near equal likelihood of starting a word. However, only a few letters have a high probability of ending a word.

Their analysis, published on Wednesday in the journal PLOS One, shows that English language words end most frequently with the letters e, s, d, t, or r, but start with a far greater diversity of letters. The study did reveal, as evident from dictionary pages, few words begin with X and Z.

"We seem to start our words with far greater freedom than the sounds with which we end our words," said Sitabhra Sinha, a senior physicist at the IMS. "There appear to be rules in what sounds can follow certain sounds."

Sinha and research scholar M. Izhar Ashraf at the B.S. Abdur Rahman University in Chennai used public databases of tens of thousands of words from multiple languages and found the same pattern, albeit at varying degrees, in every language they scrutinised.

They call it "directional asymmetry" - greater diversity of sounds at the start of words but restricted sounds at the end of words. The pattern also shows the directionality of scripts - for instance, in Arabic, Hebrew, Persian and Urdu, the greater diversity of symbols appears at the right terminal of words, less diversity at the left terminal of words.

The scientists say their analysis of inscriptions from the Indus Valley Civilisation, which still remain undeciphered, appear to corroborate earlier suggestions by archaeologists that the inscriptions were written from right to left.

"Archaeologists made the right-to-left suggestions based on looking at well-spaced symbols on the right that become increasingly cramped on the left as space for the inscription begins to run out," Sinha said.

Language specialists say the Chennai study provides fresh information about mathematical properties of languages. In the past, such studies have primarily focused on statistical analysis of how often words occur in languages.

"This is a new find that needs to be explained," Ewan Dunbar, a linguist at the University of Paris who was not associated with the study, told The Telegraph. "An exciting possibility is that this will lead us to understand something better about the human mind - how it stores words or how they are learned."