From Cuneiform Archives to Digital Libraries: The ... ?· From Cuneiform Archives to Digital Libraries:…

  • Published on

  • View

  • Download


  • From Cuneiform Archives to Digital Libraries:

    The Hermitage Museum Joins the

    Cuneiform Digital Library Initiative

    Natalia Koslova Peter Damerow

    Hermitage Museum, St. Petersburg Max Planck Institute for the History of Science,


    The Hermitage Museum in St. Petersburg wasone of the first museums to join the CuneiformDigital Library Initiative (CDLI), an opensource initiative aiming to create a digital li-brary of cuneiform documents of the 3rd mil-lennium B.C. The Hermitage collection, con-sisting of approximately 2,000 tablets of thisperiod, has been digitized and became an inte-gral part of the growing virtual CDLI library.This virtual library provides an unprecedentedmethod of accessing the oldest written sourcesof mankind. The library contains images, stan-dardized transliterations, and working envi-ronments based on language technology. Todevelop and implement a digital library of suchunusual objects challenges both the traditionalresearch standards of the highly specializeddiscipline of Assyriology and the communityof information management specialists. It re-quires an intensive cooperation between ITspecialists and scholars of cuneiform writingunparalleled in traditional research and devel-opment activities.

    1 The cooperative project

    The Cuneiform Digital Library Initiative (CDLI) repre-sents the efforts of an international group of Assyriolo-gists, museum curators and historians of science tomake freely available through the Internet images andthe contents of cuneiform tablets dating from the begin-ning of writing, ca. 3200 B.C., until the end of the 3rd

    millennium. The CDLI thus deals with the most ancientwritten documents in the history of mankind. A sub-stantial collection of about 2,000 cuneiform texts fromthis period is kept at the State Hermitage Museum in St.Petersburg. As early as June 2000 the Hermitage Mu-seum, represented by its director Prof. Michail B.Piotrovskij, agreed to the open access policy of the

    CDLI and decided to join this initiative. The digitizationof the Hermitage Museum collection became one of thefirst museum projects to be supported by the CDLI [1].

    The CDLI became the focus of a growing networkof cooperation between research institutions, universi-ties and museums. This network offers unprecedentedresearch conditions for scholars of all scientific disci-plines dealing with the cultural heritage of the ancientNear East. The CDLI is, in particular, associated withthe Philadelphia Sumerian Dictionary Project, which iscompiling the first comprehensive dictionary of theSumerian language. This dictionary will be accessiblein electronic form through the Internet. Its entries willbe linked to the body of sources provided by the CDLI.

    The CDLI was founded and is institutionally sup-ported by the University of California at Los Angelesand by the German Max Planck Society. These institu-tions will ensure the long-term availability of theCDLIs outcome until other institutions start to realizethe potential of information technology in establishing anew infrastructure for open access to the cultural heri-tage of mankind and guarantee its longevity.

    2 The objects of digitization

    2.1 Cuneiform writing

    Given the unusual nature of the objects documented inthe cuneiform digital library, it may be useful to de-scribe them in some detail before addressing the techni-calities of their digitization.

    The cuneiform script was mainly written on claytablets. It received its modern designation from thewedge-shaped stylus impressions composing its signs.Cuneiform writing was invented in the ancient NearEast at the end of the 4th millennium B.C. It was thenused for more than three millennia to write texts in al-most all the languages spoken there at that time, aboveall Sumerian and Akkadian, two major languages of theancient Mesopotamian civilization, the center of whichwas located in what is now modern Iraq. Clay is an ex-tremely durable material and for this reason an enor-mous amount of texts have survived from all thelengthy periods in which cuneiform writing was used.Proceedings of the 5

    th Russian Conference on Digital Li-braries RCDL2003, St.-Petersburg, Russia, 2003

  • 2.2 The early history of cuneiform writing

    The first documents, written on clay tablets in so-calledproto-cuneiform script, appear at the end of the 4th mil-lennium B.C. in the ancient city of Uruk, modernWarka, in the south of Mesopotamia. These texts aremainly administrative records revealing a developedsystem of retaining information and calculating goods,human and material resources.

    On the whole we have about 5,000 proto-cuneiformarchaic texts dating back to 3200-2800 B.C from differ-ent archaeological sites. After a gap of as long as twocenturies, the next period from which texts dating ap-proximately to 2600 B.C. survived is called the Fara pe-riod, relating to the modern name of the site where thefirst and very important texts of this period were found.The Fara period is represented by about 2,000 cunei-form documents now known definitively to be writtenin the Sumerian language.

    The question of whether the earlier proto-cuneiformtexts were also written by Sumerians, suggesting thatthe Sumerians invented the script in Mesopotamia, isstill under discussion. It is characteristic of this incipientproto-writing that no representation of spoken languagecould be identified so far. But in comparing the proto-cuneiform archives with those of the Fara period, wecan observe a number of continuities in the cuneiformsystem of writing and in the form and content of tabletswhich attest to the fact that at least the written traditionfound in 3rd millennium Mesopotamia was uninter-rupted.

    The history of the Mesopotamian written tradition inthe second half of the 3rd millennium B.C. is conven-tionally divided into three periods:1. Old Sumerian (ca. 2500-2400 B.C.) represented byabout 2,000 cuneiform texts mainly from the ancientSumerian city of Girsu (modern Telloh),2. Old Akkadian (ca. 2350-2200 B.C.) represented byabout 6,000 cuneiform texts from different sites writtenpartly in Sumerian and partly in the Akkadian language,3. Neo-Sumerian (ca. 2150-2000 B.C.) represented byabout 100,000 cuneiform texts from different sites,again written mainly in Sumerian.

    Although we also have school texts, so-called lexi-cal lists, from as early as the archaic period representingthe very beginning of writing, most of the cuneiformtexts of the 3rd millennium B.C. are administrativedocuments, literary texts, and royal inscriptions.

    2.3 The decline, disappearance and rediscovery ofcuneiform writing

    The last political state whose administration was con-ducted in the Sumerian language, the so-called third dy-nasty of the ancient city of Ur, collapsed around 2000B.C. After that Sumerian was gradually forced fromeveryday use by the East Semitic Akkadian, and thusbecame a dead language. The cuneiform system ofwriting, however, continued to be used in Mesopotamiaand neighboring regions and was adopted into otherlanguages until it finally died out in the first centuriesA.D. It fell into oblivion and was widely unknown in

    the Greek-Roman world, which would later determineWestern culture.

    Cuneiform writing was rediscovered in the 18th

    century by travelers who noticed inscriptions on rockfaces which survived from the Assyrian and Persianempires of the 1st millennium B.C. Attempts to deciphersuch inscriptions started immediately but failed for along time to come. Cuneiform writing was essentiallydeciphered only in the 19th and early 20th century wheninnumerable cuneiform texts were distributed through-out the world. While most of the languages of the cunei-form texts are well known today, the earliest of them,the Sumerian language of the 3rd millennium B.C., stillraises many unsolved questions. The creation of a reli-able dictionary of this language and the reconstructionof the development of the cuneiform signs from proto-cuneiform to standardized sign forms are some of themain challenges facing current studies, which call forinnovative implementations of information technology.

    2.4 The extant cuneiform sources

    The cuneiform tablets to be digitized come from ancientarchives that were either excavated by archeologists or,to a greater extent, plundered by dealers of antiquitiesand their helpers. Although the total number of thesetablets can only be estimated, it is surely in excess ofone million.

    In the current phase of the project the CDLI is lim-ited to the cuneiform archives of the 4th to 3rd millen-nium B.C., which mostly contain Sumerian administra-tive records. The number of extant tablets inscribedduring this period of early state formation amounts toapproximately 120,000 tablets and tablet fragments.These tablets were discovered in archives unearthed byvarious excavations starting in the 1880s: the Frenchexcavations in Telloh (ancient Girsu), the German ex-cavations in Warka (ancient Uruk) and Fara (ancientShuruppak), the British excavations in Tell Muqayir(ancient Ur), the British-American excavations in Jem-det Nasr (ancient name unknown), and the Americanexcavations in Abu Salabih (ancient name also un-known).

    Between the regular campaign seasons thousands ofclay tablets were plundered by local people. Some ofthe sites were excavated solely by clandestine diggerswho then sold their booty to antiquity dealers. In thisway thousands of cuneiform texts found their way intoEuropean and American museums and private collec-tions and are now dispersed throughout the world.

    2.5 The virtual reunification of ancient archives

    One of the CDLIs primary goals is to reconstruct theancient archives and join the widely dispersed butclosely related documents in a virtual library. The statearchive of the Old Sumerian Girsu, for instance, con-tained at least 1,600 tablets, which are today scattered incollections in Paris, Istanbul, Berlin, Baghdad, NewHaven, St. Petersburg, and many other places. It there-fore often happens that closely related tablets, whichwere kept in ancient times in one and the same basket,

  • are now dispersed in different places and have lost thetextual context that made the content of the individualtext understandable.

    An example may illustrate the consequences anddemonstrate the importance of easy access to differentcollections, even for studying only one particular text.Sumerian administrative documents of the third dynastyof Ur are usually dated according to special formulas atthe end of the text, for instance Year when Amar-Suenbecame king. However, there are some ambiguouscases. The formula Year when the place of Simurumwas destroyed can refer to both the 25th year of thereign of Shulgi, the second and in many respects themost prominent ruler of the dynasty, and the 3rd year ofthe reign of Ibbi-Suen, the last ruler of the dynasty.

    One of the documents in the Hermitage collectionbearing this ambiguous formula mentions an officialnamed Ur-Ishtaran who received a number of sheep,probably in order to fatten them. The same person isacting as a shepherd in several texts dating back to theperiod between the 25th and the 46th years of Shulgi.Another document bearing the same formula was sealedby an official named Ushmu. This seal is attested intexts from the 8th year of Amar-Suen, the son and suc-cessor of Shulgi, to the 2nd year of Ibbi-Suen. Accord-ingly, the first document is supposed to come from the25th year of Shulgi, the second one rather from the 3rd

    year of Ibbi-Suen. But the texts that allow for this con-clusion are now kept in the British Museum and in amuseum in Istanbul, in the universities of Yale andPrinceton, in Rome, Leiden, Geneva and many otherplaces. Using traditional methods, an enormous amountof scholarly work was necessary to identify the textsand to draw the conclusion. In contrast, the virtual li-brary built up by the CDLI makes it possible to have allthese texts on the screen within a few minutes ofsearching.

    The virtual reunification of the ancient cuneiformarchives not only facilitates philological research on thelanguages of cuneiform texts, but also makes themmeaningful for scholars of other disciplines working onsubjects related to their content, in particular for schol-ars who study the economic, social and political historyof ancient Mesopotamia. Moreover, a well-designeddigital library makes these sources useful for educa-tional purposes and interesting to the public. Digitiza-tion also protects ancient texts from the misfortunes ofmodern history. Digital images cannot substitute thereal tablets, but the extensive electronic documentationof the tablets makes each one easily identifiable. Therecent tragic plundering of the Iraq Museum wouldhave been a little less fatal had its collections alreadybeen digitized as part of an open Web of Culture rep-resenting the cultural heritage of mankind.

    3 The cuneiform collection of the Hermit-age Museum

    The Hermitage cuneiform collection contains 1,945administrative texts of the 4th to the 3rd millennium B.C.The majority of them (1,579 tablets) are Neo-Sumerian;

    the rest includes two proto-cuneiform tablets, 343 OldSumerian documents from Girsu and 21 Old Akkadiantexts from different sites. None of these tablets comefrom regular archaeological excavations.

    The first acquired cuneiform tablets were purchasedby the Royal Hermitage Museum at the end of the 19th

    century. In the museum archive there is a report fromthe chief curator of that time about the purchase in 1898of various remarkable antiquities from a well-knownFrench dealer named M. Sivadzhan. Among these an-tiquities were 78 tablets with cuneiform inscriptionsfrom Babylonia, the majority without doubt Neo-Sumerian documents. The next acquisitions were madeby the State Hermitage Museum after 1917. Some tab-lets were purchased, some of them simply confiscatedfrom private collectors.

    The main part of the Hermitage cuneiform collec-tion comes from one of Russias most famous privatecollections, that of N. Likhachev, a prominent Russianhistorian who was interested in all kinds of objects fromRussian icons to ancient manuscripts, Egyptian papyriand Babylonian clay tablets. In most cases it is almostimpossible to trace how the tablets found their way intoLikhachevs collection.

    According to some private notes in Likhachevs ar-chive he purchased Neo-Sumerian texts from Tellohmainly from Sivadzhan in Paris at the end of the 19th

    century. Some remarks in Likhachevs private corre-spondence indicate that the so-called messenger textsfrom Djokha (ancient Umma, Neo-Sumerian period),that are lists of rations for officials travelling betweenthe administrative centres, were also acquired in Parisfrom Elias Gjou between 1900 and 1914. Anothersource was Naaman in London. Since 1904 Drehem (aprominent site of the Neo-Sumerian time) was oftenmentioned in the letters of Gjou to Likhachev as theprovenance of the tablets.

    Before 1917, when Likhachev was the owner of thecollection, it was open to scholars who worked on thedecipherment of cuneiform writing. Two prominentRussian Assyriologists of that time, M. Nikolsky and V.Shilejko, worked on texts at Likhachevs residence. In1915, Nikolsky published about 1,500 texts from thiscollection.

    It is not exactly known what happened to Lik-hachevs cuneiform collection after the revolution. Atthe end of 1917 more than 1,300 tablets were moved toMoscow, among them nearly all tablets published byNikolsky in 1915. In 1918 the remaining collection be-came public property and Likhachevs home in St. Pe-tersburg was transformed into a museum under his di-rection, the State Museum of Paleography. Neverthe-less, in 1919 Likhachev was still able to sell some cu-neiform tablets to the State Department of Arts and An-tiquities, possibly the tablets that were already in Mos-cow. Many of them later found their way into the Push-kin Fine Arts Museum in Moscow.

    This State Museum of Paleography in St. Petersburgexisted until 1938. After its closure, the objects weremoved to the Hermitage. After 1938 only a few moretablets were purchased from individuals.

  • 4. The digital library

    4.1 Overview

    The digital library of the CDLI is a growing network ofdata representing cuneiform tablets, their structures,their contents, and results of scholarly work on them.The library provides a working environment, which fa-cilitates scientific work on these data. All services arefreely available on mirrored websites in Los Angelesand Berlin [2].Currently, the following facilities are provided withinthe framework of the virtual library of the CDLI:1. a comprehensive catalogue of cuneiform tablets,2. digital images of original tablets,3. scanned autographs of tablets,4. text-coding standards for cuneiform script,5. transliterations of tablets,6. tools for handling the digitized sources,7. publications with links to digitized tablets,8. show cases for specific collections,9. educational materials.In the near future, ongoing work of the project will fur-thermore provide a comprehensive list of the cuneiformsigns of the 3rd millennium B.C. documenting their de-velopment from proto-cuneiform to developed cunei-form writing. Furthermore, language technology is be-ing developed for the morphological analysis of Sume-rian and Akkadian grammatical forms in order to pro-vide translation aids and automatic linking of the trans-literations of cuneiform tablets to the forthcoming elec-tronic dictionary of the Philadelphia Sumerian Diction-ary Project [3].

    4.2 The CDLI catalog

    The CDLI has developed a format for a common cata-logue of all collections, which unifies published cata-logues into one publicly accessible database and com-pletes and improves the catalogue entries during thework on the individual collections. This catalogue,which is continuously expanded, currently contains in-formation about more than 75,000 tablets dating to the3rd millennium B.C. It contains data on the archaeologi-cal provenience and chronological classification, onphysical characteristics, the collection they belong to,and on the publication history of the registered tablets.

    4.3 Digital images of original tablets

    The display of digital images taken from the originaltablets is an essential difference of cuneiform digital li-brary to any traditional publication of cuneiform texts,which, at best, present some example photos togetherwith the representation of the tablets by drawn copiesand/or transliterations.

    Since the original tablets are valuable and usuallycannot be moved out of the collection, the images mustbe captured in the institutions where they are kept. Thisis the main reason why sophisticated technology such asthree-dimensional scanning cannot be used as long asthe required equipment is not easily transportable. The

    images are captured instead using rather common two-dimensional scanners, which are temporarily broughtinto the museums and collections by the CDLI.

    The digitization of approximately 2,000 cuneiformtablets of the Hermitage Museum pursued in the year2000 was one of the CDLIs pilot projects in the phasedeveloping the applied methods.

    Images are captured from all six sides of the tablets,usually with a resolution of 600 dpi, and stored as TIFFraw images. The post processing of these images in-cludes combining all six images of a tablet in one com-posite archival image which is the starting point for anylater processing such as enhancing the readability andconverting the images into JPG compressed formats fordisplay on the CDLI Internet websites. Before this time-consuming processing is finished, images of the obverseand the reverse of the tablets are already preliminarilydisplayed

    Currently, about 1,300 final composite images areaccessible through the Internet. Further 6,800 unproc-essed images are accessible in preliminary representa-tions of tablets. Some 2,000 further images are in theprocess of preparation for display.

    Figure 1: Drawing (2:1) of a slightly damaged cunei-form tablet [4].

  • 4.4 Digitized drawings

    Copying tablets by drawing is the standard way of theprimary publication in cuneiform studies. The drawncopies are not neutral reproductions of the tablet butrather the first step of interpretation. To draw whatseems to be visible on a tablet requires decisions aboutthe nature of any detail of the surface texture of thetablet, whether it represents a natural detail of its mate-rial surface, an effect of some kind of damage, or a traceof the impression of the stylus with which the tablet waswritten. Drawings of tablets can neither be replaced byphotos, nor by digital images. This is the reason whythe CDLI complements the scanned images of originaltablets with scans of existing drawings of them.

    Since it is technically much easier to scan drawingsthan to scan the original tablets, the number of scanneddrawings already accessible on the CDLI websites isactually much higher than the number of scans of theoriginal tablets. Currently about 32,200 composite im-ages of drawings scanned with a resolution of 150 dpican be accessed at these websites.

    4.5 The definition of a document type description foran archival XML format for cuneiform text translit-erations

    The most important interpretative step in the studyingof cuneiform documents is their transliteration. The dif-ficulty of this step results from the specific character ofthe cuneiform writing system. In principle, the cunei-form script was a syllabic writing system. The graph-emes represented certain phonemes consisting of con-sonants and vowels. But this representation was inmany respects ambiguous. Cuneiform writing involvedpolyphony and homophony, that is, most graphemeshad more than one phonetic value, and different graph-emes could represent one and the same phoneme.Moreover, the same graphemes were also used aslogograms representing entire words. They were furtherused for phonetic glosses and for determinatives. In thisfunction they did not belong to the part of the text thatrepresented spoken language. They were rather added inorder to disambiguate the use of other graphemes, eitherphonetically or semantically. Furthermore, in the 3rd

    millennium elements of two completely different lan-guages, Sumerian and Akkadian, were often simultane-ously contained in the same text. The situation is madeworse by the fact that usually the tablets show varioustypes of damage that often make parts of the text un-readable. To transliterate a cuneiform text thus includesfor each recognizable grapheme a decision about itsactual function and phonetic reading in the specificcontext is was used.

    A transliteration system able to document, on theone hand, all these decisions and, on the other hand, theactual graphemes on the tablet has to be correspond-ingly complex. Accordingly, the transliteration of cu-neiform texts is based on sophisticated traditional sys-tem of transliteration conventions using indices and dia-critics in order to distinguish homophones and specificsigns and special characters for indicating what traces

    of cuneiform signs justify the specific reading proposedby the transliteration.

    This system long resisted any attempt to be mappedinto electronic data formats. When the advantages touse computer technology became too obvious to befurther neglected, formatting capabilities of text proces-sors and specifically developed fonts with strange char-acters dominated its application for representing trans-literations of cuneiform texts. Such methods were suit-able for adapting the display and printing of electronictransliterations according to traditional conventions, butthey were of little use for any sophisticated text proc-essing using language technologies.

    Once this became obvious, attempts have been madeto simplify the system of conventions in a way that pureASCI text could represent the transliterations. This,however, made things even worse. The simplified ASCItransliterations lost so much of the precision of the tra-ditional system that they could not be used any longerfor traditional publications, nor did they provide fa-vourable conditions for the development and applica-tion of sophisticated language technologies.

    Right from the start, the CDLI has worked inten-sively on a more adequate solution to the problem ofelectronically coding cuneiform transliteration. Themajor outcome of this solution is an XML documenttype description, developed in cooperation with thePhiladelphia Sumerian Dictionary Project, which suita-bly captures the traditional transliteration conventionsand makes the often only tacitly applied rules explicit ina machine-readable form. At the same time this XMLformat standardizes the various transliteration systemsapplied by scholars according to personal and oftensomewhat idiosyncratic preferences. It makes it possibleto integrate the work on a comprehensive and reliableSumerian dictionary with the editorial work on the pri-mary sources it has to be built on.

    The XML document type description of the CDLIcovers a wide range of different aspects of cuneiformwriting such as:1. the nature of the object carrying cuneiform writing,2. the identification of the object by metadata,3. the format and subdivision of the object,4. the location of cuneiform writing on the object,5. the location of damage to the object,6. the graphemes on the object,7. the actual phonetic values of the graphemes,8. the role of graphemes in composite graphemes,9. the words containing the graphemes,10. the role of graphemes as glosses etc.,11. the numerical notations composed of graphemes,12. the actual values of the numerical notations, and13. the errors of the scribes.

    The document type description of the CDLI [5]characterizes a cuneiform text by an XML hierarchy ofabout 20 different elements and their attributes. Thetop-level element , representing a collection ofcuneiform texts, and its tree of child elements ,, , , and the line element determine unique IDs, classify the objects intotypes, and specify the location of each line of translit-

  • eration on the object which is usually a tablet, an enve-lope, or a prism. The content of the line element con-sists of parsed character data which are further struc-tured by a hierarchy of inline elements such as fornumerical notations, for words, for deter-minatives and phonetic glosses, for graphemes, for compound graphemes and so on. Furthermore,elements such as the element , the non-line element , and the non-grapheme element provide tools for standardized descriptions ofpeculiarities such as special rulings, empty space, ordamages.The transliteration of the first two lines of the cuneiformtablet depicted in figure 1 may serve as a simple exam-ple of the CDLI archival XML format:

    4(disz) gu44(disz) ab2



    2(u)5(disz) kam



    4.6 The definition of an ASCI input format for cu-neiform text transliterations

    The XML document type definition captures in a ma-chine-readable format the highly sophisticated tradi-tional conventions of cuneiform scholars. It meets therequirements for future development of language tech-nology for the Sumerian and the Akkadian languages,thus inevitably, however, becoming too complex to beused as an input format for Assyriologists transliteratingcuneiform tablets.

    For this reason the CDLI has defined a secondtransliteration format, which is an ASCI text format(ATF) that reconciles the requirements of processing

    data electronically with approved methods and conven-tions of cuneiform specialists. For all elements and at-tributes of the archival XML format, the definition ofATF contains corresponding simple, but strict and un-ambiguous rules on how to type the transliterations.These rules are derived from common habits used toexpress the complex syntax and phonetic of cuneiformwriting, replacing, however, conventions such as fontstyles and non-ASCI characters, which are incompatiblewith pure ASCI coding. Special ASCI characters, whichare not required for the transliteration itself, are used tomark information complying with elements of the ar-chival XML format. The complexity of the XML formatis reduced by suitable defaults in accordance with tacitrules of traditional transliteration conventions.

    The cuneiform tablet of figure 1 may illustrate thefeasibility of the defined format. A traditional publica-tion would render its lines in the following way:

    CMAA 002-C0005 (=P212341)obv.

    1) 4 gu4 4 ab22) e2 muhaldim3) u4 25-kam4) zi-ga5) ur`- den2-lil2-la2

    rev.1) [ ]2) (blank)3) mu-us2-sa ki-mash

    ki ba-hulThe corresponding ATF transliteration defined as ASCIstandard by the CDLI reads:

    &P212341#CMAA 002-C0005@tablet@obverse1. 4(disz) gu4 4(disz) ab22. e2 muhaldim3. u4 2(u) 5(disz)-kam4. zi-ga5. ur#- {d}en2-lil2-la2@reverse$ broken$ blank1. mu-us2-sa ki-masz{ki} ba-hul

    While this rendering is obviously close to traditionalconventions, it is entirely explicit and strictly standard-ized so that it can be automatically converted into thearchival XML format from which the ATF rules are de-rived. This conversion is realized, in cooperation withThe Archimedes Project [6], by a web service [7].The transliteration file has to be uploaded using an In-ternet browser. The file is then automatically convertedinto an XML file, which can be downloaded. This webservice is complemented with an XML validation pro-cedure, which helps to correct formatting errors of inputdata and to disambiguate and correct legacy data thathave been reformatted with errors into ATF. This vali-dation procedure is again realized by a web service [8].

  • 4.7 Computer-assisted translation and interpretation

    Cuneiform tablets document the culture of early citiesand empires in the Near East better than other knowntextual sources do for any other early civilization in theworld. The contents of these tablets range from mun-dane receipts and running accounts of a hypertrophicbureaucracy, to the incomparable verbal art of variousliterary and religious genres, to scribal exercises repre-senting the earliest manifestations of scientific thinking.Such contents are of substantial interest to scholars ofmany disciplines as well as to the public in general.

    At present however, the digital library of the CDLIprimarily addresses the demands of cuneiform special-ists. The main reason for this current bias towards theirneeds is the language barrier between modern commu-nication and the ancient cuneiform documents written inSumerian and Akkadian.

    This is particularly the case for administrativedocuments. While the literary texts aroused the interestof scholars as well as the general public early on andtranslations are widely available even through the Inter-net [9], the administrative documents have been longconsidered trivial or at least less interesting.

    The administrative texts are, in fact, philologicallysimple and, isolated from their original context, notvery informative. The translation of the same tablet de-picted in figure 1 may illustrate this:

    obv.1) 4 oxen 4 cows2) (for the) household (of the) cook,3) 25th day,4) booked out:5) Ur-Enlila

    rev.1) [ ]2) (blank)3) Year after: "Kimash was destroyed".

    The text is in many respects characteristic of ad-ministrative documents. Literary texts represent spokenlanguage. Consequently, they are philologically com-plex. Administrative documents, in contrast, representeconomic activities. They widely use technical termsand standardized designations for objects, agents, andactions. Many of these terms are not used in inflectedform. Philologically more complex phrases such as yearnames are often stereotyped. The conditions are there-fore favourable for the development of computer pro-grams, which are able to analyze administrative docu-ments and link them to dictionaries, thus making thecontent of such documents accessible also for scholarsother than cuneiform specialists.

    There are two ways to establish such language sup-port. The first is to create a program that analyses in-flected forms using grammatical rules in order to de-termine the root so that the words can automatically belinked to dictionary entries. The second is to collect theempirically inflected forms of as many words as possi-ble. Both methods complement each other. The second

    method provides the basis for the construction of effec-tive rules for programs realizing the first method. Thisis true in particular for texts such as the cuneiform ad-ministrative documents, which display specific ar-rangements of technical terms rather than typical struc-tures of spoken language. It turned out that, in fact,simpler rules apply to the formation of these texts thanto that of literary texts.

    The Hermitage collection serves as a model case forrealizing such a combined strategy. Traditional publica-tions of transliteration are often combined with care-fully produced glossaries (see e.g. [10]). The work onsuch glossaries for the publication of transliterations oftexts from the Hermitage collection is pursued within acomputer-assisted working environment, which regis-ters all individual decisions about inflected forms. Thusa data set of relations between roots, inflected formsand translations is built up which serves as a basis forthe derivation of rules for electronically classifying ad-ministrative documents and determining their content.

    4.8 The online publication of interpretations

    The digitization of cuneiform tablets and of translitera-tions of their content not only provides easier access tothese sources than traditional publications but also thepossibility of linking them electronically with interpre-tations. In order to realize this potential in an effectiveway, the CDLI has set up an electronic journal and abulletin with direct links to the digitized sources [11].The Cuneiform Digital Library Journal (CDLJ) is anon-profit, refereed electronic journal for cuneiformstudies. It seeks substantive contributions dealing withthe major themes of the Cuneiform Digital Library Ini-tiative, that is, text analyses of 4th and 3rd millenniumdocuments (incorporating text, photographs, data,drawings, interpretations), early language, writing, pa-leography, administrative history, mathematics, metrol-ogy, and the technology of modern cuneiform editing.The Cuneiform Digital Library Bulletin (CDLB) pub-lishes short notes that deal with specific topics, colla-tions, etc., and do not attempt to offer synthetic treat-ment of complex subjects.

    5. The CDLI Website

    The cuneiform digital library is accessible through awebsite at the University of California at Los Angeles,which is mirrored at the Max Planck Institute for theHistory of Science in Berlin. Like other websites of in-stitutions and projects, the website offers extensive in-formation about the project, including the aims and ac-tivities of the initiative, the associated scientists whocontribute to its work, the methods applied, reportsabout the outcome of meetings etc.

    This information about the CDLI, however, ac-counts for only a small part of the total information of-fered by the CDLI website. The sites main function isto provide free access to the growing number of cata-logue data, images and transliterations, which constitutethe library, and to assure the longevity of this access

  • with stable links to the sources beyond the time per-spective of current CDLI activities.

    In its present state, the site already shows substantialcomplexity. Nevertheless, it is still under construction.On the one hand, the number of registered tablets aswell as the amount and quality of the data representingthem are continuously increased. On the other hand, thefacilities of the working environment it represents areenriched and new levels and functions are implemented.

    The core of the digital library access system is thedisplay of data concerning an individual cuneiformtablet or other object with cuneiform text. The standarddisplay offers essential catalogue data, composite im-ages of the six sides of the tablets displayed to scalewith regard to their real sizes, the transliterations oftheir cuneiform text content, and links to further infor-mation such as full catalogue entries, 600 dpi high-resolution images, or images of hand copies. The dis-play can be toggled between two layout versions, onefor optimal screen display, the other for printing.

    Moreover, as a service to collections digitized by theproject, the CDLI offers the opportunity to have theircollection additionally displayed using templates, whichpreserve their corporate identity (see the entry page tothe collections [12]). These collection-specific displayenvironments also provide particular access facilitiessupporting the administration of the collections

    Access to the display of individual cuneiformdocuments is offered by search facilities and by cata-logue listings of individual collections. The preliminarysearch environment, which is presently implemented,allows searches in essential fields of the CDLI cata-logue and for text searches in the transliterations.Search results can be displayed in different ways. Theycan be accessed by simple lists of links to the standarddisplay, by lists containing essential catalogue data andimages, or by browsing the standard display through thesearch results.

    The search facilities will be continually expanded inthe future so that other types of data such as extendedsets of metadata, glossaries, paleographic differentia-tions etc. can be included into search requests.

    At the beginning, the CDLI followed a simple strat-egy to achieve a robust display of the data. A PERLscript was used to create a system of static HTMLpages. This strategy made it possible to distribute thedata not only through the Internet but also by means ofstorage media such as compact discs.

    The increased volume of data hosted by the CDLIforced the project to give up this simple strategy in fa-vour of a dynamic solution. At present, a Filemaker Proweb server provides the data for the standard display ofthe cuneiform tablets, essentially using three files withdata, the CDLI catalogue file, another file containingthe transliterations in a display format generated by aconversion program from the ATF transliterations, anda file which contains the URLs to related high resolu-tion images for each cuneiform text.

    The following list of presently implemented unitsgives an overview of the current state of the CDLI web-site:1. the CDLI home site with static pages containing in-

    formation about the project,2. the standard display system of the digital library,3. the search unit of the digital library (preliminary

    version),4. specific display systems for individual collections

    (to be continually extended),5. access pages to philological tools (preliminary ver-

    sion),6. access pages to technical tools and services (to be

    continually extended),7. access pages to the Journal CDLJ and the Bulletin

    CDLB,8. educational pages (to be extended).


    [1] The Hermitage Cuneiform Collection.

    [2] Cuneiform Digital Library Initiative. A joint pro-ject of the University of California at Los Angelesand the Max Planck Institute for the History of Sci-ence in Berlin. site:

    [3] The Pennsylvania Sumerian Dictionary.

    [4] R. K. Englund. The Ur III Collection of theCMAA. In CDLJ, 2002.001.

    [5] DTD for CDLI XML.

    [6] The Archimedes Project Realizing the Vision ofan Open Digital Research Library for the Study ofLong-term Developments in the History of Me-chanics.

    [7] ATF to XML converter

    [8] SGML Parser

    [9] The Electronic Text Corpus of Sumerian Literature.

    [10] N. Koslova. Ur III-Texte der St. Petersburger Ere-mitage. Wiesbaden, Harrasowitz, 2000.

    [11] CDLI Publications.

    [12] CDLI Digital Library

    The CDLI is funded by the Digital Libraries Initiative(NSF/NEH), United States, and the Max Planck Society,Germany.


View more >