Physicists are not the only researchers to look for patterns in data in an effort to find underlying structure – linguists also need to find patterns, and they have had a recent hand from the concept of entropy. Rob Lee of the University of Exeter and colleagues compared symbols inscribed by the Picts – a Scottish Iron Age culture dating back to about the 4th to 9th centuries AD – with texts known to be in a written language (though not Pictish).

Measures of repetitiveness as quantified by entropy showed the inscriptions to be much closer to a modern language, with a small vocabulary, than to simple pictorial (no pun intended!) scripts. The work is particularly interesting because it could be used even when samples of whatever symbols are written are small, as is the case for known examples of Pictish. This method could also be extended to see how much linguistic content animal communications, such as those of dolphins, contain.