Title: Constructing word similarities in Meroitic as an aid to decipherment
ArXiv ID: 0808.3616
Date: 2009-08-24
Authors: Researchers from original ArXiv paper
📝 Abstract
Meroitic is the still undeciphered language of the ancient civilization of Kush. Over the years, various techniques for decipherment such as finding a bilingual text or cognates from modern or other ancient languages in the Sudan and surrounding areas has not been successful. Using techniques borrowed from information theory and natural language statistics, similar words are paired and attempts are made to use currently defined words to extract at least partial meaning from unknown words.
💡 Deep Analysis
Deep Dive into Constructing word similarities in Meroitic as an aid to decipherment.
Meroitic is the still undeciphered language of the ancient civilization of Kush. Over the years, various techniques for decipherment such as finding a bilingual text or cognates from modern or other ancient languages in the Sudan and surrounding areas has not been successful. Using techniques borrowed from information theory and natural language statistics, similar words are paired and attempts are made to use currently defined words to extract at least partial meaning from unknown words.
📄 Full Content
This paper addresses a technique using a combination of known words and techniques from information theory to try to decipher the meanings of additional words in the extinct and undeciphered language, Meroitic. First, I will give a short history of the language and the problems translating it and next describe the statistical techniques and their results and implications.
1 A Short History of Meroitic (Török, 1997;Lobban, 2004) Meroitic was the written language of the ancient civilization of Kush, located for centuries in what is now the Northern Sudan. The word ‘Meroitic’ derives from the name of the city Mero, which was located on the East bank of the Nile south of where the Atbara River flows off to the east. It is the second oldest written language in Africa after Egyptian hieroglyphs. It is a phonetic language with both a hieroglyph form using some adopted Egyptian hieroglyphs and a cursive form similar to Egyptian Demotic writing. The language had one innovation uncommon in ancient written languages such as Egyptian hieroglyphics or Greek in that there was a word separator, similar in function to spaces in modern scripts, that looks similar to a colon (see Figure 1). Meroitic was employed starting the 2nd century BC and was continuously used until the fall of Mero in the mid 4th century AD.
The script was rediscovered in the 19th and 20th centuries as Western archaeologists began investigating the ancient ruins in the Sudan. The first substantial progress in deciphering Meroitic came around 1909 when British archaeologist Francis Llewellyn Griffith was able to use a bark stand which had the names of Meroitic rulers in Meroitic and Egyptian hieroglyphs. The Meroitic hieroglyphs were then corresponded to the Meroitic cursive script and it was then possible to transliterate Meroitic (see Figure 1). Some vocabulary was later deciphered by scholars including loan words from Egyptian, gods, names, honorifics, and common items. However, the language remains largely undeciphered. The greatest hope for decipherment, a Rosetta stone type of tablet containing writing in Meroitic and a known language such as Egyptian, Greek, Latin, or Axumite, has yet to be found. Further confounding research is the confusion regarding which language family Meroitic belongs to. Cognate analysis has proceeded extremely slowly since it is disputed to which language family Meroitic properly belongs. Recent work by (Rilly, 2004) has suggested that Meroitic belongs to the North Eastern Sudanic family, however, full decipherment is still elusive.
Meroitic was one of the earliest ancient languages to be investigated using computers (Leclant, 1978;Heyler, 1970Heyler, , 1974;;Ouellette, 1999). Much of this work was dedicated to creating an alphabetical index of Meroitic and also comparing Meroitic words to possible cognates in Nubian or other known ancient and modern languages from the region. In (Smith, 2007), many of the longest texts were analyzed by ranking words according to frequencies to verify whether the current texts we have follow the mathematical relation Zipf’s Law where the word frequencies f vary with the rank z according to the relation
where α ≈ 1. In analyzing the Meroitic texts, though many did not fit the strict criterion of α ≈ 1, the frequency-rank distribution followed the behavior of a truncated power law distribution whose exact parameters varied by text. Some texts such as the long stela REM 1003 (REM is a text designation that stands for Répertoire d’épigraphie méroïtique, the most comprehensive catalogue of Meroitic texts) more closely fit Zipf’s Law though. From these results, without knowing the meaning of the text it is clear that the statistical variations and occurrences of words in the Meroitic texts in our possession are not surprising and mirror those of other human languages. Though this may seem a trivial property at first glance, it gives us the hope of using more advanced statistical Taken from the latest font set for Meroitic Hieroglyphic and Cursive characters developed by the Meroitic scholars Claude Carrier, Claude Rilly, Aminata Sackho-Autissier, and Olivier Cabon. Web Address: http://www.egypt.edu/etaussi/informatique/meroitique/meroitique01.htm
3 techniques to help tease some of the meaning from the unknown portions of the language.
At the outset, I acknowledge that no language has ever been fully deciphered using purely statistical or mathematical techniques and I am not proposing that Meroitic will be completely understood using these tools. In particular, many of the subtleties of human semantics and syntax are irregular or do not follow a consistent pattern that statistics is usually excellent at analyzing. What this paper will attempt to do is not claim to derive the meaning, a loaded concept in the study of linguistics, of a word but rather find words which are used very similarly in the text. When two words are used very similarly with one of the words being known, we can hope to
