Language evidence in Indus script

Read more below

By G.S. MUDUR in Delhi
  • Published 23.04.09

New Delhi, April 23: Scientists have obtained what may be the first mathematical evidence to support the idea that the 4500-year-old inscriptions from the Indus civilisation were graphic representations of a language.

A research team from India and the US has found that certain statistical features observed in the Indus inscriptions are similar to those in several other natural languages — ancient and modern.

The scientists from the Tata Institute of Fundamental Research, Mumbai, and other institutions have shown that the Indus script had patterns of flexibility also found in Sumerian, Old Tamil, English, and Sanskrit. Their research shows that this flexibility is absent in sequences of non-linguistic systems such as human DNA, proteins, or the artificially-created computer language, FORTRAN.

“We can now say with more confidence than ever before that the Indus writings encoded a language — even though we still don’t know what that language might be,” Ronojoy Adhikari, a team member and physicist at the Institute of Mathematical Sciences, Chennai, said.

The new findings will appear in the US journal Science tomorrow.

The Indus script is etched on small seals, tablets, and amulets found over the past several decades at sites across northwestern India and eastern Pakistan where the Indus civilisation flourished from around 2500 BC to 1900 BC.

The script uses only 417 symbols or signs, and inscriptions are short with only 4 or 5 signs. But the script has defied all attempts at decipherment. The new research used computer science techniques to track statistical features shared by languages.

“This is the first quantitative evidence to show that the Indus script represents a language, thereby implying that the Indus civilisation was literate,” said Rajesh Rao, associate professor of computer sciences at the University of Washington, Seattle, and lead author of the study.

A leading historian said the new work used a novel technique to bolster an old idea.

“This work uses a different method to strengthen a well-accepted idea that the Indus script encoded a language,” said Nayanjot Lahiri, professor of history at the University of Delhi, and an authority on the Indus civilisation.

“Almost everyone has worked with the assumption that it is a language. The issue has been which language — and that remains unsolved,” Lahiri told The Telegraph.

The scientists worked with the principle that symbols in spoken languages do not occur at random after each other, nor is there a rigid ordering of symbols.

There is some amount of flexibility in choosing the next symbol or letter. For instance, in English, the letter “t” may be followed by vowels such as “a” or “e” or “i” and some consonants such as “r”, but typically not by “d” or “f” or “x”.

In a project supported by the Sir Jamsetji Tata Trust and the US-based Packard Foundation, the researchers computed this flexibility in the Indus script, and in languages as well as in non-linguistic sequences — DNA, proteins, and FORTRAN.

“The Indus inscriptions fell right in the middle of the spoken languages,” said Mayank Vahia, an astronomer at the TIFR, Mumbai, and principal investigator of the project initiated about three years ago.

The scientists said they now planned to focus on the structure of the script.

“We’d like to determine the grammatical rules the script followed, and compare the structure of this script with other writing systems to gain more insights about the content,” said Nisha Yadav at the TIFR.

But the new research has also drawn sharp criticism from three experts who have argued that the Indus script had no linguistic basis but was merely a collection of symbols and pictograms to depict religious or political symbols.

In a controversial paper five years ago — titled “The Collapse of the Indus-Script Thesis: The Myth of a Literate Harappan Civilisation” — Michael Witzel, professor of Sanskrit and Indian Studies at Harvard University and two others — Steve Farmer and Richard Sproat — had put forward several powerful arguments to claim that the Indus civilisation lacked a true script and may have even rejected writing.

Witzel and his colleagues point out that the Indus inscriptions are too short to encode messages. The average size is 4 or 5 symbols, the longest has 17. They have questioned the methodology used in the new study.