Making a Stone Speak: A Computational Reconstruction of the Missing Text of the Singapore Stone

The Singapore Stone was a sandstone epigraph inscribed with around fifty lines of text in a writing system which has not be found anywhere else in the world, ‘hiding’ an unknown language. The stone was once located at the mouth of the Singapore River, at the so-called ‘Rocky Point’. In 1843, the Singapore Stone was blown up by the British, to open the access to the mouth of the river and to build new facilities. Only three fragments were recovered by Lt. Col. James Low and sent to Calcutta, to be studied. Presumably, only one of the fragments was given back to the Lion City and is currently preserved at the local National Museum.

So far, the text has eluded all attempts at deciphering it, both at the level of analysis of the writing system and at the level of its reading. Over time, Researchers have proposed several possible writing systems considered compatible with the characters of the Singapore Stone and a number of languages possibly ‘hidden’ behind them. All decipherment attempts have been unsuccessful.

Our study aims to use methods from the fields of Computer Vision, Artificial Neural Networks, and Deep Learning to reconstruct the missing parts of the inscription of the Singapore Stone, starting from the fragment we have and extending the investigation, potentially, to the whole epigraph. The above-mentioned methods are implemented by using both Python and OpenCV, while digital images of the epigraphic document are enhanced via image-processing methods, such as image restoration, thresholding, and edge detection, in order to identify feature points.

The use of deep-learning methods to study the inscription of the Singapore Stone and to reconstruct the missing parts of its text will provide the scientific community with a new method to enhance the understanding of its writing system and with an innovative contribution to the decipherment efforts of this enigmatic document, by developing an original approach in Computational Epigraphy.

Bio:

Dr Francesco PERONO CACCIAFOCO (Ph.D. University of Pisa, Italy) is, currently, a Senior Lecturer in Historical Linguistics at the Linguistics and Multilingual Studies Programme (LMS), School of Humanities (SoH). An Etymologist by training, he works mainly on the etymological reconstruction of Indo-European place names, on the study of Aegean Civilizations and Scripts, on the documentation of languages from South-East Asia belonging, in particular, to the Papuan and Austronesian families, and on the History of Cryptography and Crypto-linguistics. Francesco has been unsuccessfully working, since 1999, on the deciphering of Linear A, an undeciphered Aegean writing system from Crete, used in the Bronze Age and ‘hiding’ the so-called and unknown Minoan language. He is the co-author of the forthcoming book Place Names: Approaches and Perspectives in Toponymy and Toponomastics published by Cambridge University Press. He lives with his wife and two cats, and every day he learns something valuable from them.