Reading Akkadian cuneiform using natural language processing

Abstract

In this paper we present a new method for automatic transliteration and segmentation of Unicode cuneiform glyphs using Natural Language Processing (NLP) techniques. Cuneiform is one of the earliest known writing system in the world, which documents millennia of human civilizations in the ancient Near East. Hundreds of thousands of cuneiform texts were found in the nineteenth and twentieth centuries CE, most of which are written in Akkadian. However, there are still tens of thousands of texts to be published. We use models based on machine learning algorithms such as recurrent neural networks (RNN) with an accuracy reaching up to 97% for automatically transliterating and segmenting standard Unicode cuneiform glyphs into words. Therefore, our method and results form a major step towards creating a human-machine interface for creating digitized editions. Our code, Akkademia, is made publicly available for use via a web application, a python package, and a github repository.

Fig 1. Example of cuneiform writing from a gold tablet of Ashurnaṣirpal II.

This is a foundation inscription buried under the palace of Ashurnaṣirpal in the city Apqu (modern Tell Abu Marya, see link [3]), where it was found millennia later by archaeologists. After giving the titles of the king, the text introduces itself as a foundation inscription, written on gold and silver (the silver parallel of this copy was found alongside the gold). This is an anomaly for cuneiform texts, which were usually written on clay. Therefore, these inscriptions are a symbol of great wealth and extravagance. It ends with curse formulae for later princes who may dare to erase Ashurnaṣirpal’s name from his inscriptions. It is currently in the Yale Babylonian collection (YBC 2398; YPM BC 16991), republished from the Yale Peabody Museum of Natural History under a CC BY license, with permission from Klaus Wagensonner, original copyright 2020.

Fig 2. Example of Unicode cuneiform.

On the left are the original cuneiform characters written on the obverse of the gold tablet introduced in Fig 1. On the right is the Unicode version which was generated from the transliteration of this inscription.

Fig 3. An illustration of the bidirectional LSTM.

Fig 4. Output of the algorithms.

These are lines 31-34 of the second column of Sennacherib’s clay prism, probably from Nineveh, now in the Israel Museum (IMJ 71.072.0249). The text records eight campaigns of the Assyrian King, including the siege of Jerusalem which is well known from the Book of Kings. The line reads: ‘On my return march, I received a heavy tribute from the distant Medes, of whose land none of the kings, my ancestors, had heard mention.’ (translation adapted from A.K. Grayson and J. Novotny’s edition available on ORACC, Q003497).