The Idea of Writing: Writing Across Borders 9789004215450, 9789004217003

Table of contents :
Contents
Acknowledgements
Invention and Borrowing in the Development and Dispersal of Writing Systems
27–30–22–26 – How Many Letters Needs an Alphabet? The Case of Semitic
Nubian Graffiti Messages and the History of Writing in the Sudanese Nile Basin
About “Short” Names of Letters
Early Adaptations of the Korean Script to Render Foreign Languages
Han’gŭl Reform Movement in the Twentieth Century: Roman Pressure on Korean Writing
The Character of the Indian Kharosthī Script and the “Sanskrit Revolution”: A Writing System Between Identity and Assimilation
Symmetry and Asymmetry, Chinese Writing in Japan: The Case of Kojiki (712)
Writing Semitic with Cuneiform Script. The Interaction of Sumerian and Akkadian Orthography in the Second Half of the Third Millennium BC
Old Wine in New Wineskins? How to Write Classical Egyptian Rituals in More Modern Writing Systems
Subject Index
Language (Group) and Script Index
Author Index

Citation preview

The Idea of Writing

The Idea of Writing Writing Across Borders

Edited by

Alex de Voogt and Joachim Friedrich Quack

LEIDEN • BOSTON 2012

Cover illustration: Multi-lingual sign posted by the Greek-Norwegian mission on Sai Island, Sudan, using Arabic and Old Nubian scripts for the Nubian language. This book is printed on acid-free paper. Library of Congress Cataloging-in-Publication Data The idea of writing : writing across borders / edited by Alex de Voogt and Joachim Friedrich Quack.   p. cm.  Includes index.  ISBN 978-90-04-21545-0 (alk. paper)  1. Writing. 2. Written communication. 3. Language and languages—Orthography and spelling. I. Voogt, Alexander J. de. II. Quack, Joachim Friedrich, 1966–  P211.I34 2011  302.2’244—dc23 2011037983

This publication has been typeset in the multilingual “Brill” typeface. With over 5,100 characters covering Latin, IPA, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities. For more information, please see www.brill.nl/brill-typeface. ISBN 978 90 04 21545 0 (hardback) ISBN 978 90 04 21700 3 (e-book) Copyright 2012 by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Global Oriental, Hotei Publishing, IDC Publishers, Martinus Nijhofff Publishers and VSP. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change.

CONTENTS Acknowledgements  ........................................................................................

vii

Invention and Borrowing in the Development and Dispersal of Writing Systems .......................................................................................... Alex de Voogt

1

27–30–22–26 – How Many Letters Needs an Alphabet? The Case of Semitic ...................................................................................................... Reinhard G. Lehmann

11

Nubian Grafffijiti Messages and the History of Writing in the Sudanese Nile Basin  .................................................................................. Alex de Voogt & Hans-Jörg Döhla

53

About “Short” Names of Letters  ................................................................. Konstantin Pozdniakov

69

Early Adaptations of the Korean Script to Render Foreign Languages  ..................................................................................................... Sven Osterkamp

83

Han’gŭl Reform Movement in the Twentieth Century: Roman Pressure on Korean Writing  ................................................................... Thorsten Traulsen

103

The Character of the Indian Kharoṣṭhī Script and the “Sanskrit Revolution”: A Writing System Between Identity and Assimilation  ................................................................................................. Ingo Strauch Symmetry and Asymmetry, Chinese Writing in Japan: The Case of Kojiki (712)  .............................................................................................. Aldo Tollini

131

169

vi

contents

Writing Semitic with Cuneiform Script. The Interaction of Sumerian and Akkadian Orthography in the Second Half of the Third Millennium BC  ........................................................................ Theo J.H. Krispijn

181

Old Wine in New Wineskins? How to Write Classical Egyptian Rituals in More Modern Writing Systems  .......................................... Joachim Quack

219

Subject Index .................................................................................................... Language (Group) and Script Index  ......................................................... Author Index  ....................................................................................................

245 249 251

ACKNOWLEDGEMENTS The Cluster of Excellence “Asia and Europe in a Global Context: Shifting Asymmetries in Cultural Flows” at Heidelberg University hosted the exchange of thoughts and ideas that were at the basis of this publication. In particular, we wish to thank Profs. Madeleine Herren-Oesch, Axel Michaels and Rudolf Wagner for their support. The Idea of Writing seminar series started at Leiden University where it was hosted for fijive years. This is the fijirst volume that resulted from a meeting outside Leiden, at Heidelberg University in Germany. Subsequent seminars have been held in Einsiedeln, near Zurich, Switzerland and Venice, Italy. They illustrate the lasting interest that this exchange has created. We owe particular thanks again to Connie Dickmeyer for her corrections and suggestions, and the patient stafff at Brill Publications who have facilitated the continued productivity of these exchanges by publishing this second volume of contributions. The publication of this volume has been made possible by generous funding of the Deutsche Forschungsgemeinschaft (German Research Foundation) in the context of the cluster of Excellence “Asia and Europe in a Global Context”.

INVENTION AND BORROWING IN THE DEVELOPMENT AND DISPERSAL OF WRITING SYSTEMS Alex de Voogt In the history and dispersal of writing systems, borrowing is a central feature. Only in three or perhaps four known instances, script was invented independent of contact with other writing or prior knowledge of writing. Mesopotamian, perhaps Egyptian, Chinese and at least one MesoAmerican writing system are thought to have developed independently while all subsequent systems borrowed at least the idea of writing. Exposure to writing is not sufffijicient to develop a writing system for a language. Today speakers of a language may be literate only in a language other than their own. The ability to write down one language does not sufffijice for writing down another. For this a series of inventions or borrowings are necessary. They include a set of signs and a set of sign values. These choices also determine whether a script is considered alphabetic, syllabic, logographic or a variation or combination of these. Once a system is in place, the agreement among the users of the script will determine the development and future of its use. The above scenario suggests that all elements of a script are necessary before writing can occur. In other words, apart from the independently invented scripts mentioned above, there is no gradual development towards writing but only a sudden appearance of a ‘complete’ script that subsequently may develop further. This is particularly true for scripts that borrowed the idea of writing since a growing awareness with incremental steps towards the idea of writing, which one could consider an evolutionary development, is not necessary. The process of signs changing into lexical items, rebus-forming and the slow development of writing is only attested, or perhaps only necessary, if the idea of writing was not present from the start. In all other instances the development of a script is an exercise that can be completed by one person, such as for Cherokee, or decreed by an administration, as is suggested for the cuneiform script used for Old Persian, or at least can be established in a relatively short time and with limited means. Even the presence of a complex society, an elaborate administrative system within a state or other elements that point at a practical need for writing are not necessarily at the basis of

2

alex de voogt

later script development. Later scripts, i.e., scripts that were created after the idea of writing was already present in a region, are unlikely to use signs for administrative purposes fijirst and then develop these into script signs over time since they are already aware of the use of signs for writing. Rather, if the idea of writing is known—in some cases people may even be literate in another language—, the script signs and values are borrowed or invented, and the entire writing system is created relatively sudden. Previous essays on the history and development of writing systems suggest that a category of independently invented scripts exists separate from the above-mentioned systems in Mesopotamia, China and Meso-America. Examples of these scripts often have a known inventor. Daniels (2007: 55; Daniels & Bright 1996: 579) termed them unsophisticated grammatogenies when the inventor is unfamiliar with the workings of an existing writing system. The unsophisticated grammatogenies, such as, according to Daniels, the Bamum and the Caroline Islands scripts, always appear to have a syllabic system of writing. It is argued here that invention takes place on multiple fronts and if the idea of writing is already present, it is better to speak of borrowing and innovation processes. The Caroline Islands script came about after a failed introduction of an alphabet by a stranded missionary. The otherwise illiterate islanders then developed a syllabary for which some of the sign shapes were clearly borrowed (Riesenberg & Kaneshiro 1960, de Voogt 2010a). The processes that involve the development of a script are much diffferent from one example to the other. Most of these so-called invented scripts, whether the grammatogeny was sophisticated or not, are only partly invented and difffer widely in their borrowing strategies. Borrowing after the Creation of a Script After the creation of a script a separate process of script development takes place. Signs can change shape, gradually instead of abruptly, over time. In the case of writing loanwords, the users of a script can be in need of signs with diffferent sound values than those present in their script. Even though diffferent writing systems have co-existed since antiquity, they rarely borrow each other’s sign shapes once the script has been created, particularly if the scripts are not related. Cuneiform signs do not appear in Egyptian and the English alphabet has not introduced Cyrillic signs to introduce the odd Russian loanword. This is diffferent for numeral systems, diacritics and special signs such as question marks or dollar signs.

the development and dispersal of writing systems

3

These types of signs have been introduced into diffferent alphabets as well as, for instance, into Chinese and Japanese writing. In contrast, when a new script is developed, the inventors may take sign shapes from several diffferent scripts and incorporate them into their sign list. Cherokee has Cyrillic and Roman alphabetic signs, and Maldivian Thaana has Arabic numerals that were transformed into consonant signs (DeSilva & Sugathapala 1969, de Voogt 2010b). These examples, however, are not without complication. In the latter case, it is not clear where many of the other consonantal signs were taken from. In the case of Cherokee, its syllabary signs were partly transformed when they were used for printing purposes, several years after the inventor of the syllabary, Sequoya, successful introduced the script to other Cherokee people. It appears that part of the inventory developed by Sequoyah was at least inspired by existing alphabetic signs, while another part was somewhat similar but was made to resemble alphabetic signs more emphatically when they were printed and, fijinally, a small portion underwent signifijicant changes to make them look like the known English alphabet (Cushman 2010). Many signs remain unique to the Cherokee syllabary but this borrowing and redesign process for, at least, part of its inventory has required detailed historical analyses before it became clear (Cushman 2010; Walker 1984, 1985; Walker & Sarbaugh 1993). As soon as a writing system is in place, it is unlikely that sign shapes from yet other writing systems are borrowed into the system. Instead new signs with shapes similar to the ones already present are much preferred. In sum, the sudden creation of such a script is followed by a gradual development of that script. The Possibilities of Borrowing Today, the most common creation of a script is the application of an alphabet to a previously unwritten language, and often some new signs or new values of existing signs are developed before the writing system can be put to use. In the development of a new writing system where the idea of writing is already present, the signs, the value of the signs and the resulting system of writing are borrowed in degrees and the previously mentioned category that includes, for example, the Caroline Islands script, Cherokee and Old Persian, does not prove to be an exception. The following examples illustrate the possibilities of borrowing. These possibilities can be summarized as follows:

4

alex de voogt Sign shapes are borrowed or partly borrowed from another script. Sign values are borrowed or partly borrowed from another script, i.e., the sign shape is not always borrowed with its accompanying sign value. The system of writing is similar or diffferent from the borrowed script.

It is possible to borrow from more than one script and both ancient and modern examples exist. With each script from which a new system has borrowed, the above three elements can be distinguished. If placed in a table, the three elements of borrowing create the following borrowing scenarios:

Sign shape Sign value System

L1

L2

L3

L4

L5

L6

L7

No No No

Yes Yes Yes

Yes Yes No

Yes No Yes

Yes No No

No Yes Yes/No

No No Yes

Language 1 (L1) borrowed no elements and this happened in the three regions mentioned above that created the Cuneiform, Chinese and MesoAmerican scripts. It is also possible for other scripts in later times to fall into this category but they would need to have had a new set of sign shapes unrelated to those found anywhere else. In the history of writing systems, this seems rarely the case. Possible contenders are Linear B from ancient Crete, Rongorongo from Easter Island or the earliest Egyptian writing. In these cases, it remains difffijicult to show that all signs were indeed invented and that the system of writing, syllabic or logographic, was not borrowed. It sufffijices to state that the (un)sophisticated grammatogenies also do not fall into this category. L2 borrowed all elements and much of the alphabetizing projects of today are of this nature. As was mentioned above, some parts of the script may be borrowed rather than the entire script and this scale makes the above table only a fijirst and crude categorization. A number of the (un)sophisticated grammatogenies fall into the category of L3 in which sign shapes and values are borrowed but the system of writing is diffferent, usually syllabic or alpha-syllabic. Meroitic writing is an example from antiquity in which a series of signs was taken from Demotic and Egyptian hieroglyphs for both the cursive and hieroglyphic variation. The sign values were mostly the same but the system changed from the complex Egyptian system (with coexisting phonetic and logographic writing, plus the semantic marking of categories) to the Meroitic

the development and dispersal of writing systems

5

alpha-syllabic system. The basic factor of the change in this case was not so much active addition but the abandoning of much of the model. L3 is a borrowing category for which many examples exist but this is not the case for L4. Here the signs and the system are borrowed but the sign values have been changed. In the case of Maldivian Thaana, Arabic numeral signs were changed to consonantal signs. Arabic is an abjad system, while Thaana has alpha-syllabic characteristics (see de Voogt 2010b/c). Carian shows an L4 borrowing strategy with alphabetic signs that were taken from Old Greek but for which almost all sound values were changed. For instance, the delta received an /l/ sound, gamma represented /b/ and mu became /s/. It is said that this was a conscious efffort to distinguish its writing system from Greek (Adiego 2007). L5 may apply to much of the printed form of Cherokee where the signs appear borrowed from existing alphabetic signs but the sign values are diffferent and the system is syllabic rather than alphabetic. It is noted that the strategies of L3, L4 and L5 are all diffferent with respect to the extent of borrowing that took place and the number of inventions that were necessary. They do not resemble the application of an existing script to a new language but they are independent effforts to create a script using ideas and inventions already present in neighboring areas. L6 is only a theoretical category since it suggests that the signs were invented but the values of such signs were borrowed. For instance, a secret script, as developed in war-times, in which each sign is replaced by a number, would fall into this category. The inventory remains the same but the sign shapes are changed. It may be possible to change the system of writing as well, but, to my knowledge, L6 is a rare or even unknown borrowing strategy in the history of writing. Creating a new sign also means that an accompanying sign value needs to be created. Most scripts have used a combination of the above borrowing strategies and creating such new signs is one of them and is presented here as category L7. In this category all signs and their sign values are invented. Ogham writers, for instance, invented all of their signs but kept the alphabetic system they had been exposed to. Even systems such as Braille and Morse code can be put into this category.

6

alex de voogt Other Categories of Borrowing

In addition to sign shape, sign value and system of writing, it is possible to add the borrowing or invention of numeral systems, punctuation, the direction of writing, or even script materials. They have been excluded in the list above. Their presence or choice rarely influences a sign inventory that is necessary for a writing system. The use of numeral systems has a history separate from writing systems but also seems less varied in sign shapes. Punctuation is often borrowed but since these markings can be added or eliminated without afffecting the sounds that can be represented, they are better treated in a separate study. The direction of writing also does not afffect the sound inventory represented by the signs in the script. Although writing from bottom to top, as attested in some Philippine scripts (Conklin 1949), is particularly rare, sign inventories are not necessarily dependent on the writing direction. The direction is often a preference of the users. Script materials afffect the shape of signs and in the Philippine case also partly explain the direction of writing. Their use of bamboo shoots makes writers carve away from their body so that the script is written from bottom to top. It also makes left-handed writers carve the signs in mirrorimage. Readers of the script are described as being flexible in reading texts from diffferent angles or directions with a common practice to read the signs from bottom to top. For Western alphabets, fonts are developed that are more suitable for computer screens than the earlier fonts developed for newspapers and books. Several scripts show separate inventories of signs for writing in stone or on parchment, paper and other soft materials. This often doubles the number of sign shapes and such versions may be treated as separate scripts in terms of borrowing strategies. In other cases, the signs are adapted to be used on a diffferent writing material but their borrowing strategy may remain the same. The Development of Script In sum, contact with writing may transfer the idea of writing but the creation of a script also requires sign shapes and values that combine into a writing system. This latter development consists of borrowings and inventions that need to be largely completed before the script can be put to use. This part of script development is not gradual but abrupt. As soon as a fijirst sign inventory is put in place, the script may develop further and add or change sign shapes and values. These, mostly gradual,

the development and dispersal of writing systems

7

transformations are also influenced by changes in the language and the use of other writing materials. At this stage new sign shapes are less commonly borrowed from other scripts. The examples from Mesopotamia, Meso-America and China where script developed independently have a preceding stage of script development in which the idea of writing emerges gradually. Examples of administrative systems that develop into writing suggest such a gradual development. Daniels (2007) found that syllabaries are much more common for grammatogenies. However, his unsophisticated grammatogenies are not an unusual category in the light of the above. Still, it is reasonable to claim that (alpha-)syllabic systems are much more common in general than logographic scripts. To this one may add that the alphabet has become a dominant system only with the implementation of alphabetic writing to unwritten languages in the twentieth century. Writing across Borders The authors in this volume illustrate the complexity of invention, borrowing and innovation but also go much beyond the simple categories described above. They do so from many diffferent perspectives and in varying depths. This depth is determined by the available scholarship rather than the ability of the individual scholars. Together, they bring detail to a topic that is complex and layered and that gains much from a multi-varied approach. Each writing system discussed here has a history of scholarship that, in many cases, is part of a diffferent and largely separated fijield of study. Only occasional conferences can bring experts of diffferent writing systems together to enter a conversation about their development where philological and linguistic approaches allow a common language. This volume, as the previous one (de Voogt & Finkel 2010), is the result of such an interaction. No attempt has been made to force these scholars into one tradition of scholarship. Instead, there is a common focus on understanding writing systems rather than on furthering knowledge of a region or a historical period of a society. For some this allows for an abundance of references to ongoing debates, for others this is a fijirst exploration. This variance of scale points at the stage and the history of debates for each of the topics raised in this volume. The alphabet, alpha-syllabary or abugida, and logographic scripts are all represented but in diffferent historical contexts. They show the forces of

8

alex de voogt

change and the intricacies of these changes. The alphabet and its Semitic versions, often not considered alphabets but abjads, are part of a discussion of the historic and linguistic developments in the Near East. The vowels signs were already present in Semitic scripts but the Greek alphabet is often seen as the fijirst complete alphabetic system. Sign values are central in the debate that traces these borrowings at a crucial point in the history of the alphabet. The Nubians have a long history of writing but the modern language does not have an orthography. Grafffijiti writing shows a particular but widely used attempt at writing Nubian with Arabic script. This recent borrowing and adaptation of a script shows a dynamic that is best understood in light of the limited purpose of the texts. The modern alphabet has names for its letters that are borrowings with a separate history. An exploration of these names of letters shows a series of curious inventions and borrowings, an element in the study of writing systems that is particularly intriguing. The discussion of Korean as it was adapted by a bureau for interpreters, shows a rich number of strategies to represent a range of other, often unrelated, languages. New signs and adaptations of signs with new values are present, but here one can also borrow diacritics. The sign may stay the same when only part of the sign, the diacritic, is taken from another script. A second study on Korean presents the forces of Western Roman alphabets on the reform movement of the Korean writing system. Cursive versions, capitals, handwriting and printed writing, and many other forms create their own problems of developing a script that links to all the modern needs of writing. The abugida or alpha-syllabary is discussed with its earliest examples found in India. The introduction of Sanskrit as a dominant religious language is shown to explain the demise of the Kharoṣṭhī script in favor of the Brāhmī script. These abugida scripts accommodated a new language through adaptations of sign shapes and new values of signs but also strategies to represent sounds with combinations of signs, a possibility that is not necessarily a borrowing strategy but still an important way to represent foreign sounds. Japanese serves as an example of syllabic writing but the influence of Chinese writing on its system complicates the situation. It is shown how Chinese characters were used for both phonetic and logographic purposes in ancient Japan and thereby added to the complexities of their use.

the development and dispersal of writing systems

9

Cuneiform script was used for a series of languages and the early interaction of Sumerian and Akkadian orthographies presents a historically important phase in the history of cuneiform. The complex script has a history that is as difffijicult to interpret as the writing itself. Finally, unetymological writing in Egyptian demotic is an adaptation not so much to a new language but to the pronunciation of an older stage of the language that contains lexemes and morphemes for which no demotic spelling existed. A solution had to be found which could either use one-consonantal signs or groups for similar-sounding words. The strategies of writing are explained in the context of demotic writing. The diffferences of writing Egyptian by way of hieroglyphs, demotic or hieratic allow further insight in a script that together with cuneiform had a long history of development. The contributions cover both modern and ancient developments. The scripts are alphabetic, syllabic, logographic or combinations and variations of these. Most importantly, they show that borrowing, innovation and invention is of all times and may occur in any script. The understanding of these developments requires the realization that borrowing processes and invention consist of many diffferent but related parts. In the end, the versatility of those who use writing systems have created a plethora of solutions for writing languages of diffferent kinds under the varied influence of existing scripts. References Adiego, I.J. 2007. The Carian language. Leiden: Brill Publications. Conklin, Harold C. 1949. Bamboo literacy on Mindoro. Pacifijic Discovery 2/4:4–11. Cushman, Ellen. 2010. The Cherokee syllabary from script to print. Ethnohistory 57(4): 625–650. Daniels, Peter T. 2007. Littera ex occidente: Toward a Functional History of Writing. In Cynthia L. Miller (ed.), Studies in Semitic and Afroasiatic Linguistics Presented to Gene B. Gragg, pp. 53–68. Studies in Ancient and Oriental Civilization, 60. Chicago: Oriental Institute of the University of Chicago. Daniels, Peter T. & W. Bright. 1996. The World’s Writing Systems. New York & Oxford: Oxford University Press. de Voogt, Alex. 2010a. The Caroline Islands script: a linguistic confrontation. In Alex de Voogt & Irving Finkel 2010. The Idea of Writing: play and complexity, pp. 327–343. Leiden: Brill Publications. ——. 2010b. Languages and scripts in the Maldive Islands: coding and encoding. In Alex de Voogt & Irving Finkel 2010. The Idea of Writing: play and complexity, pp. 197–205. Leiden: Brill Publications. ——. 2010c. The Meroitic script and the understanding of alpha-syllabic writing. Bulletin of soas 73(1):101–105.

10

alex de voogt

de Voogt, Alex & Irving Finkel. 2010. The Idea of Writing: play and complexity. Leiden: Brill Publications. DeSilva, M.W. Sugathapala. 1969. The phonological efffijiciency of the Maldivian writing system. Anthropological Linguistics 11 (7): 199–208. Riesenberg, S.H. & S. Kaneshiro. 1960. A Caroline Islands Script. In Anthropological Papers, 60. Washington: U.S. Government Printing Offfijice. Walker, Willard. 1984. The design of native literacy programs and how literacy came to the Cherokees. Anthropological Linguistics 26(2): 161–169. ——. 1985. The roles of Samual A. Worcester and Elias Boudinot in the emergence of a printed Cherokee syllabic literature. International Journal of American Linguistics 51(4): 610–612. Walker, Willard & James Sarbaugh. 1993. The early history of the Cherokee syllabary. Ethnohistory 40(1): 70–94.

27–30–22–26 – HOW MANY LETTERS NEEDS AN ALPHABET? THE CASE OF SEMITIC* Reinhard G. Lehmann Introduction and General Considerations In his recent novel “Das Geheimnis des Kalligraphen”, the Syro-German writer Rafijik Schami reports a fijictitious dialogue between the progressive young calligrapher Hamid Farsi and his master Serani. In the course of this dialogue, the master states: Die arabische Sprache hat nur neunundzwanzig Buchstaben, und je mehr du davon zerstörst, umso unsicherer und ungenauer wird die Sprache. (Schami 2008: 395)1

Modern oriental scholars know that the Arabic alphabet has 28 letters. A twenty-ninth letter is added by tradition, attributed to Mohammed himself, and is nothing but the Lam-Alif ligature LA, which simply combines the two letters into a single cohesive unit. Accordingly, Hamid Farsi considers the Lam-Alif ligature superfluous, artifijicial, and not a genuine letter in the Arabic alphabet. However, the more conservative and careful calligraphy master Serani warns his young, impatient pupil not to be all too optimistic, modernistic, or straightforward: Ich will dich nicht entmutigen. Diesen Buchstaben LA hat der Prophet dem Alphabet hinzugefügt, und er bleibt, bis die Erde untergeht. Wenn du auf mich hören willst, streiche keinen einzigen Buchstaben, weil sonst die ganze islamische Welt gegen dich sein wird, denn diese Buchstaben kommen im Koran vor. (Schami 2008: 395)2

* Thanks to Robert M. Kerr, Philip C. Schmitz and other colleagues and friends with whom I could discuss parts of this paper. However, mistakes and errors are exclusively my responsibility. 1  The Arabic language has only twenty-nine letters, and the more of them you destroy, the more uncertain and imprecise the language becomes. (translation by R.G.L.). 2  I don’t want to discourage you. The Prophet himself added the letter LA to the alphabet, and it will persist until the world ends. If you listen to me, then you won’t do away with one single letter, or otherwise the whole of the Islamic world will disapprove of you because these letters are found in the Koran. (translation by R.G.L.).

12

reinhard g. lehmann

And: Man nimmt keinen Buchstaben weg [. . .] den die Jahrtausende geformt haben . . . (Schami 2008: 396)3

The dialogue scene points to some problems, which are directly relevant to the title of this paper: – To determine how many letters, or graphemes, an alphabet needs, it is crucial to know what a letter is. Is the Lam-Alif ligature of the Arabic alphabet a letter, or is it not? Moreover, if Lam-Alif is not a letter, why then should OMEGA, the last letter of the Greek alphabet, be labelled a letter? Yet, it is conspicuous that even the name O-mega tells us that it is nothing more than a diacritical variation to denote the ‘big’—that is, long—pronunciation of the {o} of the Greek alphabet, which thus became the Ο-mikron. This is a fact that is also proven occasionally by its early archaic written forms.4 And if the Arabic LA were a letter, why then isn’t the German digraph {CH} or the German {SZ} ligature (ß) a letter? – There is a claim that reducing the number of letters of an alphabet causes uncertainty and imponderability of the language. Is this claim really convincing? – There is the claim that a letter, once invented, will last for eternity: “You must not remove a single character [. . .], which has been formed by millennia.” This, however, is obviously not true. In the course of history, several letters have become obsolete and were removed in several alphabets. Despite the traditional Muslim’s claim that the Arabic alphabet has 29 letters, modern oriental scholars know that it actually has only 28 letters. And some might even say that the Arabic alphabet has only 15 or

3  You must not remove a single character [. . .], which has been formed by millennia. (translation by R.G.L.). 4  In areas like Ionia, Knidos, Paros, Melos and others, the diffference between the short and long /o/ was marked by use of “a new sign which appears to be a doublet formed from O by breaking the circle” (Jefffery 1961: 37), i.e., opening it at the bottom. Another interesting specimen is found in an archaic inscription from Phleious (fijirst half of the sixth century BCE) where in the word λωι the OMEGA is written with a digraph formed by two full circle dotted OMIKRONs that are written one below the other (Jefffery 1961: pl. 24,1b, also Jefffery 1961: 147 note 1). On this basis, on the other hand, Bernal even concludes that the double-circle was its most pristine shape (Bernal 1987: 15; 1990: 121).

27–30–22–26 – how many letters needs an alphabet?

13

18 letters, because some similar letters have merged graphically, such as ra with zay, ḥa with ḫa and ǧim, ba with ta and ṯa (and, the fijinal shape aside, also with ya and nun), and fa with qāf, or were secondarily diffferentiated for phonemic reasons, such as dal and ḏal, and ͑ayn and ġayn, and were thenceforward distinguishable only by additive points (Gruendler 2006: 151). These distinguishing points, or diacritic marks, have not been there from the beginning of the Arabic script, and are not in the earliest versions of the Qur’an (Gruendler 1993: 131ss). But others may argue that in earlier times, when the Arabic script evolved from an older, most presumably Nabataean or Nabataean-like ancestor, these letters of course were distinguishable, and only later merged into one common shape, thus making it necessary to add diacritical marks, or points (Gruendler 2006). This, however, implies that older Arabic or pre-Arabic texts for instance remained readable for some time anyway, even after letter shapes became very similar or merged totally (Kerr 2010: 372). It also implies that neither the merger or reduction of a graphemic inventory nor its enhancement by diacritics had its most important reason in simple readability. Thus, how many letters does the Arabic alphabet, for instance, have? And how many letters does an alphabet need at all? Anyway, the quest for the number of letters that form an alphabet could indeed be a bit complicated. The International Phonetic Alphabet (IPA) contains more than 130 graphemes, or signs, and depending on what you really count, there are even more than 160 signs, or ‘letters’. However, no one would reasonably say that the IPA is a real natural alphabet. There is presumably no individual that will actually be able to produce in speech all the sounds of this set of signs, or feels the need to do so, simply because there is not a single language that has all of these phonemes. Thus, there is a remarkable restriction: an alphabet represents only the sounds of a single natural language (or group of natural languages), which is used by a certain group of speakers at a certain time in history. Does it represent all of the sounds of such a natural language? Presumably not—there is sufffijicient evidence to assume that every natural alphabet as a graphemic system is to a certain extent defijicient in relation to the phonemic system of the language or dialect it is used for. Moreover, obviously the ratio between a phonemic and a graphemic system can change in the course of history. I will suggest in this paper that the Northwest Semitic shortened or ‘short Abgad’ does not reflect a spoken dialect (the Phoenician, as is mostly believed) at all, but has emerged as a scriptio franca for the Semiticspeaking Levant. In the late second millennium BCE, the short Abgad,

14

reinhard g. lehmann

which eventually made the game in alphabet history, provided sufffijicient consonantal contrast to constitute a to some extent supraphonemic ‘writing across the borders’ that covers a broad variety of Levantine Northwest Semitic dialects. To get this point, a Semitic alphabet typology, comments on several questions and unsolved problems, and a critical evaluation of terminology are required. Historical (and Typological) Background There is no doubt that what we call an alphabet today had developed more than 3000 years ago somewhere in Syro-Palestine, or the Levant. However, a closer look at these ancient Levantine alphabetic ideas shows conspicuously that already in antiquity there were two competing alphabetical sequences: Halaḥama and Abgad.5 Both seem most probably to have displayed close and presumably almost full match of both, the existing phoneme inventory of the language and the provided graphemes, and, what is more, both seem to have originally reflected a common phonetic inventory, albeit in a diffferent order. Of course, because they did not represent vowels, some would argue that they both are not true alphabets. Actually they really were. We will come back to this point again later. Old South Arabic – Sabaean (Sabaic): 29 What has been traditionally called the ‘South Arabic’ sequence, or alphabetic order, also called Halaḥama according to its fijirst four letters h–l–ḥ–m, is almost exclusively attested in the South-Arabic language area. From several fijinds of mainly Sabaic abecedaries, this fijixed alphabetic Halaḥama letter sequence is known already from the fijirst millennium BCE as follows: h–l–ḥ–m–q–w–š–r–b–t–s–k–n–ḫ–ṣ–ś–f– –͗ –͑ ḍ–g–d–ġ–ṭ–z–ḏ–y–ṯ–ẓ Minor deviations from this sequence are only found in some stone inscriptions, but not so far in the so-called minuscule script, which was used on

5  Of course, there are also other ‘alphabetic’ sequences of other scripts in antiquitiy, the most interesting of which is that one used by the older runic script, called the Futhark (fuþark). Theo Vennemann in several publications recently tried to derive it from the Phoenician-Punic Abgad-sequence, which in my opinion is the most convincing theory up to now (Vennemann 2006, 2009).

27–30–22–26 – how many letters needs an alphabet?

15

wooden sticks, esp. palm-leaf stalks.6 With 29 phonemes, the old South Arabic alphabet seems to preserve most completely the original Semitic consonantal phonemic inventory (Stein 2003: 18).7 Based on palaeographical criteria, the Old South Arabic script would seem to date to at least the mid-second millennium BCE. This presumably also applies to the Halaḥama sequence. This script was continued independently in Ethiopia where a South-Semitic language (Ge’ez) was adopted early in the Common era. The Ethiopic derivative of the ‘South Semitic’ writing system in a separate and isolated development due to phonetic mergers often confuses some graphemes in writing. It was also secondarily transformed into a ‘semisyllabary’ (Diringer 1968) by altering the shapes of the consonantal graphemes depending on the following vowel. Hence, one can claim that the classical Ethiopic script has either 26 consonantal graphemes or is a syllabary with 251 characters (33 × 7 + 20; cf. Getatchew Haile 1996: 573). Nonetheless, the Ethiopic script still “has the analytic depth of an alphabet” (Coulmas 1989: 153). However, despite the almost exclusive South Arabian attestation of the Halaḥama sequence and its seemingly exclusive later South Semitic history, it is noteworthy that the very earliest evidence of Halaḥama is found in the thirteenth century BCE long alphabet of 27 signs from Ugarit in Syria (see below).8 Accordingly, the so-called ‘South Semitic’ alphabet is

6  Stein (2003: 11, 2010: 591); for further references Stein (2010: 591 and 743f). Abecedaries written in stone are published by: Bron & Robin (1974), Irvine & Beeston (1988), Ryckmans (1985). The sole published deviation in minuscule script, with the interchange of ḏ and z in Oost.Inst. 37, most probably is nothing but a (mistaken) exception (Stein 2003: 11 Fn. 64). 7  A remarkable phenomenon regarding the number of letters is that only in minuscule writing, i.e. on wooden sticks, the South Arabic alphabet uses the grapheme ‹ḍ›, where for etymological reasons ‹ẓ› is to be expected. But in terms of lexicography, there are no exceptions (Stein 2003: 27f, Nebes & Stein 2008, and recently Stein 2010: 502f). 8  A reverberation of the Halaḥama is also found in a late Egyptian alphabetic device, as recently is shown by Quack 1993, 2003. Moreover there is the claim that an underlying notion of Halaḥama were also perceivable in the Latin word elementum (“letter / character as basic constituent of speech”), reflecting ha–la–ḥa–ma (Müller 1994: 309, Gruendler 2006: 148). At any rate, this would mean that the so far only rarely attested northern, or Levantine, history of the Halaḥama should not have come to an end so early after all. However, though for phonological reasons it is most probable that Latin elementum is a foreign loan word (“Herkunft unklar, jedenfalls nicht Erbwort”, Walde & Hofmann 1938: 398), Müller’s assumption has severe historical difffijiculties. If Latin elementum depends on ha–la–ḥa–ma, it would be the only fijirst millennium BCE remnant of the Halaḥama in the Mediterranean except for the above-mentioned Egyptian and a few Ancient North Arabic ‘abecedary’ inscriptions in diffferent, if any, order including a modifijied Abgad as well as Halaḥama (Macdonald 1986, 2008a: 185), which then must have been responsible for the transmission onto Latin. Yet, also the older hypothesis that elementum represents

16

reinhard g. lehmann

by no means exclusively South, but also Levantine! This is not even ruled out by an occasionally reported assumption that the Ugaritic Halaḥama were only the attempt of an Arabian tourist scribe to learn the Ugaritic script—which is nothing but an over-romanticized attempt by modern scholars to escape the aporia of the facts. The Ugaritic Evidence: 30 (27) It is a matter of fact that the earliest well attested rich inventory alphabetic system with a considerable number of texts known hitherto is the Ugaritic cuneiform, or wedge alphabet. The texts and abecedaries that have been found in Ugarit and its environs witness to three diffferent systems of a cuneiform alphabet. Currently, eighteen Ugaritic cuneiform abecedaries, combined textabecedaries or abecedary fragments are known (Puech 1986, Hawley 2008). Nearly all of these abecedaries represent the dextrograde Ugaritic long alphabet of 30 signs. Its original 27 signs were most probably adapted from an unattested Northwest Semitic linear alphabet tradition (Lundin 1987b, Dietrich & Loretz 1988). Three additional signs were doubtlessly secondarily appended hence fijinally producing a 30-letter sequence. As nearly all of these abecedaries9 show, they basically arrange the characters according to an expanded scale of the so-called Abgad device, which has its name from the fijirst four letters –͗ b–g–d of the short standard scale (see below): –͗ b–g–ḫ–d–h–w–z–ḥ–ṭ–y–k–š–l–m–ḏ–n–ẓ–s– –͑ p–ṣ–q–r–ṯ–ġ–t– i͗ – u͗ –s̀

the beginning of the second half of the Abgad order (el-em-en- . . .) in analogy to ‘Alphabet’ (Walde & Hofmann 1938: 398) fails because such half-cutting division of the alphabet is not attested in Latin, but in Semitic only (which can be seen by the Atbash encoding device, or, for instance, Psalms 9 and 10, which subdivide a totally complete alphabetic acrostic in its middle, and by Nahum 1, which covers only the fijirst half of the alphabet. See for a recent discussion Renz 2009). Moreover, the pronunciation of the letters L, M and N as ‘el’, ‘em’, and ‘en’ seems to be not much earlier than the fourth century CE. Thus, there seems to be an interdisciplinary circular reasoning, and regarding the etymology of elementum Latin philology seems to be as helpless as it was one hundred years before (thanks to Christian Tornau, Würzburg, with whom I was able to discuss this point). 9  These abecedaries are listed in Hawley 2008 and in KTU2 5.4–5.6, 5.9, 5.12–5.14, 5.16– 5.17, 5.19–5.21, 5.30. Only in the fragments RS 15.071 (KTU2 5.8), 5.274 (7.54), 12.019 (5.5), and 19.174[4] (9.342) is no decision possible whether they contained 30 or only 27 letters. The fragmentary abecedary RS 16.265 covers the missing letters in additional enigmatic (exercise) letter sequences that remind of a similar phenomenon in lines 1–4 of the proto-Hebrew 22-letter abecedary ostracon from ͑Izbet Ṣarṭah. For RS 88.2218 (dextrograde Halaḥama sequence), see below. For an overview and discussion of Ugaritic abecedaries see Pardee (2007).

27–30–22–26 – how many letters needs an alphabet?

17

This fijirst category, the cuneiform long alphabet, is by far the most widely used, represented in numerous administrative and literary texts from the north Syrian coastal sites of Ugarit and Ras Ibn Hani. Noteworthy, however, is the occasional interchange of signs of mostly similar phonemic value within the Ugaritic Long Alphabet, thus ‹t› for ‹ṣ›, as in art for arṣ (afffricate pronounciation), ‹w› for ‹h›, as in bwtm for bhtm, ‹h› for ‹ḫ› as in phr instead of pḫr, and ‹ ›͗ for ‹ ͑› as tṯar instead of tṯ r͑ (Tropper 2000: 21.352.5). Maybe such examples are more than mistakes only, and reflect vernacular variations that might have rendered similar phonemes with a reduced set of graphemes. Another long but non-expanded cuneiform alphabet of 27 signs is also attested at Ugarit, but only in one abecedary that is written dextrograde (RS 88.2215 = KTU 9.426, Bordreuil & Pardee 1995, Röllig 1998, Bordreuil & Pardee 2001). It omits at least the last three letters of the Ugaritic Standard long Alphabet and displays some more peculiarities in both letter shape and adjustment. A similar abecedary was discovered in Beth Shemesh to the west of Jerusalem, but is written sinistrograde (KTU 5.24 = 8.1, Lundin 1987, Pardee 2003/04: 18 n. 43). Both, though not entirely identical, bear witness to the so-called South Semitic alphabetic Halaḥama device.10 However, to date no Ugaritic texts are known, that are unambiguously written in this 27-letter alphabetic system. Finally, at Ugarit there was a possibly somewhat later cuneiform short alphabet system of presumably 22 letters, which could be written in either direction (Dietrich & Loretz 1988: 145–275).11 Texts written in this alphabetical system seem to be dated not earlier than the thirteenth century BCE, and are not only found three or four times in Ugarit.12 However, they also spread south from Ugarit across the Levant, to several sites

10  “Enfijin, les formes des signes de cet abécédaire si particulier sont tout aussi étrangères à Ougarit que l’ordre: aucun autre texte ougaritique n’utilise cet inventaire graphique” (Hawley 2008: 225), similarly Robin (2008: 233). However, Röllig (1998: 87) states “An ugaritischen Schreibgewohnheiten ändert dieses Alphabet ja nichts, denn der Phonembestand blieb derselbe. Man wird also in der ugaritischen Überlieferung vergeblich nach Texten im ‘südsemitschen’ Alphabet suchen.” According to Röllig, the few difffering sign forms could be blamed on a scribal scholar and are not evidence enough for use or non-use of the Halaḥama series in Ugarit. 11  As yet, there are only 21 letters of the Abgad device identifijied, missing the ‹ṣ› sign. Also possible is that there was a short alphabet tradition with 24 letters, missing only the interdentals of the long alphabet (Tropper 2000: 76–77). A table of the sign forms, which vary to a certain extend, is found in Tropper (2000: 75). 12  Ugarit: KTU 4.31 (right to left), KTU 4.710, and Minet el-Beida: KTU 1.77 (right to left). The attribution of a further text, the votive clay nail KTU 7.60 to this group is questionable, Dietrich & Loretz (1988: 168–70; 1989: 107).

18

reinhard g. lehmann

in Syria-Palestine and Cyprus.13 Some of them might also have 24 letters, what could be counted as a transitional stage from the Long to the Short Alphabet.14 Nevertheless, they bear witness to a short(ened) alphabetic cuneiform tradition. Often this short cuneiform alphabet is considered the clay adaptation of the Phoenician linear alphabet into wedge script (Pardee 2008: 7), which implies that the Phoenician alphabet was an entirely independent development (however, the question remains why there should have been such radical reduction in the phoneme inventory only some 200 km south in ‘Phoenicia’). In this Ugaritic short alphabet, there is only one letter for alef (i.e., without vowel-connotation), the ‹ḫ› sign is replaced by the ‹ḥ› sign, the ‹ġ› sign is replaced by the ‹ ͑› sign (which indicates a merger of the two sounds—or graphemes—respectively). Also the interdental phonemes /ṯ/, /ẓ/ and /ḏ/ have no distinctive grapheme but seem to be represented by their sibilant equivalents, i.e., /š/, /ṣ/, and /z/ respectively (Dietrich & Loretz 1989: 107f, Tropper 2000: 22.5). The ‘short’ alphabet was used to write texts; no abecedary is yet attested in this alphabet, and even the few examples of these texts found at Ugarit were all from outside the palace area, thus suggesting that they do not originate from the ‘offfijicial’ schools of Ugarit (Dietrich & Loretz 1989: 108). Hence, we do not know exactly whether the Ugaritic short alphabet bears witness to the Abgad sequence (though this is not unlikely), and it is uncertain whether the texts constitute a ‘phoenicianizing’ representation of the Ugaritic language or are in a kind of early plain Phoenician or another North Canaanite dialect (Pardee 2003/04:17). Anyway, lexicon and grammar of the short alphabetic cuneiform texts seem to be closer to later Phoenician or another North Canaanite dialect than to the Ugaritic texts of the long alphabet (Tropper 2000: 22.81fff; Dietrich & Loretz 1989: 110). With the destruction of Ugarit at the beginning of the twelfth century BCE the Ugaritic cuneiform alphabet tradition came to a sudden end. But

13

 Listed, for instance, by Millard (2007: 85), which are: Silver Bowl from Hala Sultan Tekke (Cyprus) 6.68; storage jar from Tell Nebi Mend/Qadesh (but with the ‹ġ› sign, thus maybe long alphabet tradition); two vessel inscriptions from Kamid el Lodz (one dextrograde, the other sinistrograde); an inscribed jar handle from Sarafand-Sarepta [Lebanon] KTU 6.70 (dextrograde, relative z and verb p l͑ indicating ‘Phoenician’ Canaanite language: Smith (2005); Vita (2003: 401f ); a bronze tablet (knife blade?) from Tabor valley (dextrograde) KTU 6.1; a small tablet from Tel Taanak (left-to-right) KTU 4.767. Dietrich & Loretz (1989: 109): “This group of texts attests inscriptions written in both directions. The widely held notion, that the short alphabet is characterized by right-to-left writing cannot be maintained. In these texts the direction of writing appears to be rather a function of local school traditions.” 14  KTU 6.68 and KTU 5.24 including the ‹ḥ› and ‹ġ› sign (Tropper 2000: 22.45).

27–30–22–26 – how many letters needs an alphabet?

19

the Abgad order that formed its dominant arrangement device (but most likely was not an Ugaritic invention, nor was alphabetic writing at all, Pardee 2007: 182) survived as the basic order of most Western and Semitic alphabets until today. The arrangement of the short alphabet in a 22-letter Abgad device as it is attested from the late second millennium BCE onwards seems to be nothing but an abbreviation, or selection from the (Ugaritic) long Abgad series. Its most common standard device, now starting plainly with –͗ b– g–d, is known from several Northwest Semitic abecedaries of the fijirst millennium BCE, but best from the Hebrew Bible:15 –͗ b–g–d–h–w–z–ḥ–ṭ–y–k–l–m–n–s– –͑ p–ṣ–q–r–š–t The as yet oldest authority for a 22-letter short Abgad sequence, however, is the proto-Hebrew linear exercise on the I͑ zbet Ṣarṭah sherd, dated— by archaeological context data—to the late twelfth century. In its bottom line, it clearly displays a genuine Abgad device, although with slight alterations (Cross 1980; Hallo 2004: 285f): –͗ b–g–d–h–m–ḥ–z–ṭ–y–k–l–[ ]–n–s–p– –ṣ–q–q–š–t A quite similar arrangement is found in the late 10th century Tel Zayit stone abecedary (Tappy et al. 2006; Sanders 2008; Rollston 2008b), which was discovered in 2005: –͗ b–g–d–w–h–ḥ–z–ṭ–y–l–k–m–n–s–p– –ṣ–q–r–š–t While the apparent position change of the bilabial sounds /w/ and /m/ in the ͑Izbet Ṣarṭah sherd most probably had phonemic reasons, the ‹ḥ›–‹z› and the ‹ ͑›–‹p› interchanges here (and in a few other abecedaries) are remarkable. The latter is also found in some other proto-Hebrew (and Hebrewonly, or Southern Levantine, but not Ugaritic!) abecedaries of the early

15  Unfortunately, the numerous Northwest Semitic linear abecedaries are mostly either fragmentary or represent a selection of some letters only. However, these all incontrovertibly bear witness to the Abgad series or at least segments thereof. In the Hebrew Bible, though, there is a considerable number of complete abecedaries attested in the following Biblical Hebrew acrostics, where every colon, verse, or stanza starts with a new letter of the alphabet: Psalms 9-10 (disturbed between Lamed and Pe), 25 (Waw missing), 34 (Waw missing), 37, 111, 112, 119, 145 (Nun missing), Prov 31:10-31, Lamentations 1, 2, 3, 4, and Sirach 51:13–30 (Zayin-Kaf missing), Nah 1:2-8 (fijirst half), outside the Bible also in the Dead Sea Psalms scroll 11Qpsa xxi:11–17, xxii:1–15, xxiv:3–17. For Biblical Hebrew acrostics in general, see Freedman (1999: 1-24), for abecedaries, most recently Gzella (forthcoming).

20

reinhard g. lehmann

fijirst millennium, the most prominent of which are the abecedaries 1 and 3 from Kuntillet Adjrud in the southern Negev.16 Moreover, this very peculiarity of an ‹ ͑›–‹p› to ‹p›–‹ ͑› interchange is also found in some Biblical acrostic poems.17 Though, in any case, this deviation from the majority or ‘standard’ sequence does not have a phonemic reason, it is not likely that it was just haphazard or mistaken. Rather, its long afterlife even in the Hebrew Bible seems to trace back to a diffferent Abgad variety, which must be seen in connection with local or areal ‘Hebrew’ scribal traditions. Thus, to conclude the fijirst overview, there seem to have been fijive main classes of early alphabet records, to be categorized by graphic format (cuneiform or linear), number of graphemes, abecedary-device, and direction of writing:

16  This ‹p›–‹ ›͑ modifijied Abgad is also found 3 times on an unprovenanced early sixth century BCE Hebrew ostracon of the Moussaiefff collection, which was recently published by Martin Heide (Heide 2007). 17  Among the aforementioned (n. 15), a modifijied order with interchange of ‹ ›͑ –‹p› to ‹p›–‹ ›͑ is found in Psalms *9–*10 and Lamentations 2, 3 and 4, and also in the assumed underlying Hebrew text of the Greek (Septuagint) version of Prov. 31. That this ‹p›–‹ ›͑ modifijied Abgad device is as yet found outside the Hebrew Bible in South Levantine or ‘proto’-Hebrew abecedaries only, is maybe of some importance. Anyway, the majority of later Abgad abecedaries in general bears witness to the ‹ ›͑ –‹p› order, which accordingly we may label the ‘standard’. It is not only found in most Biblical poetic acrostics (see note 15), but in an early ostracon of the eighth century from Lachish, too. Unfortunately though, most Hebrew Abgad abecedary inscriptions and also some Phoenician, Aramaic, and Ammonite abecedary sherds or seals are either fragmentary or too short respectively to reveal whether they testify to a ‹ ›͑ –‹p› or to a ‹p›–‹ ›͑ order. Accordingly, nothing can be said about which of both variants was the most common. Quite diffferently, Sanders (2008: 102, referring to Byrne 2007: 4–5) claimed the ‹p›–‹ ›͑ sequence to be the older one because of the well known Hebrew scribal habit of placing a ‹p› directly under a ‘superscript’ ‹s›, which, in his opinion, “could well have been triggered” as a “ghostly trace of a diffferent order entirely” by the Halaḥama device with its ‹s›–‹p›–‹ ›͗ –‹ ›͑ sequence. However, Byrne’s underlying notion is not convincing. Abecedaries, whatever their main function might have been, were not very useful as a curricular guide (as Byrne 2007: 5 claims) “to gain familiarity and profijiciency with the characters, their relative size, and their relative placement” (Byrne 2007: 5). Would it not have been much more efffective to gain scribal profijiciency in baseline control, relative letter size, and letter placement by copying texts (not abecedaries), for these alone provide most or all of the combining possibilities? Beyond sole reading and writing knowledge, mastership-level scribal profijiciency is fijirst and foremost knowledge of layout and kerning devices (Lehmann 2008b), which were acquired by practicing, writing and copying texts again and again. Moreover, Sanders obviously ignores the fact that the ‹p›–‹ ›͑ sequence, which he claims to be “found in no Israelite abecedary”, is well attested at least three times in the Hebrew Bible in reasonably late texts (Lamentations 2–4), which alone rules out any idea that this order might be the earlier, Canaanite, and non-Israelite.

27–30–22–26 – how many letters needs an alphabet?

21

– Cuneiform (wedge) long alphabet with expanded inventory (30): Abgad device, left-to-right (dextrograde) – Cuneiform (wedge) alphabet with full inventory (27): Halaḥama device, dextrograde and sinistrograde (so far only two examples) – Cuneiform (wedge) alphabet with reduced inventory (22): Abgad device? (no abecedary), sinistrograde (right-to-left) – Linear alphabet with full inventory (27): Abgad device? (no abecedary), sinistrograde – Linear alphabet with reduced inventory (22): Abgad device, sinistrograde To be honest—there is a ‘fake’, or misleading assumption, in this overview because there actually is no full-inventory 27-sign linear alphabet as yet found. This is only what scholars normally expect to have been there.18 Nevertheless it is not attested. Maybe there are some non-linear, quasipictographic forerunners of the rich-inventory alphabet in the early second millennium, depending on how you interpret the protosinaitic and related inscriptions of the second millennium BCE. Nonetheless, they are not linear. As yet, the full inventory 27-sign linear alphabet is a phantom, as is a yet-unattested 27-sign cuneiform (wedge) Abgad sequence. However, while there are good reasons to assume the existence of the latter,19

18  See, for instance, the entirely useless efffort made by Martin (1962) to identify in the sub-text of the so-called Rapa’ palimpsest arrowhead a linear ‹ḏ›-letter, bearing the shape of ‹z› “with two small slanting strokes appended to the upper crossbar” (p. 183). All necessary objections to this is said by Cross (1967: 14* n. 34). Relying strictly on the claim already made by W.F. Albright that there must have been a Northwest-Semitic linear long 27letter-alphabet, as far as I know, Martin was the last who seriously tried to fijind something like that in any Northwest Semitic linear alphabetic inscription. It is obvious that Martin failed totally. As a side-efffect of his fatal methodological meander, unfortunately also other useful and valuable insights of Martin remained underestimated. 19  The fijinal position of the surplus letters i͗ , u͗ , s̀ in the expanded cuneiform Abgad points clearly to a primal 27-letter Abgad that closed with the letter ‹t›. Originally, there was one -͗ sign only, which was a pure consonantal glottal stop (the a͗ sign). The vowelspecializing signs i͗ and u͗ , as well as s̀ predominantly for loanwords, were added at a later stage (Sivan 2001: 9–10). Furthermore, the abecedary RS 23.492 (= KTU 5.19) has an unexpected and otherwise inexplicable dividing stroke after the ‹t›-sign, which confijirms that the users of the 30-letter cuneiform Abgad by all means were aware of the secondary character of the last three signs i͗ , u͗ and s̀ (Bordreuil 1982: 9–10: “. . . confijirme que les utilisateurs de l’alphabet ougaritique de trente lettres étaient conscients du charactère supplémentaire des trois dernières lettres i, u, et s̀.”). However, the KTU edition as well as Tropper (2000: 22) interpret the divider as a missing part of the i͗ sign, which, from a palaeographic point of view, is less probable.

22

reinhard g. lehmann

it remains unlikely that there ever was something like a linear long Abgad. Thus, the fourth class in the above list should be disregarded. As for the remaining classes, there are two basic parameters: the arrangement of letters, and the number of graphemes. It seems as if here lies the historical clue for the question under discussion. Beforehand, however, a clarifijication of terminology is necessary. Abjad and Alphabeta—the ‘Daniels’ Distinction Among the four abovementioned parameters, graphic format and direction-of-writing are predominantly dependent on extrinsic factors such as areal economic conditions and availability of writing material, or sociological and regional trends of scribal behaviour.20 Thus, apparently it seems most important to keep apart the number of graphemes and the abecedary-device, which are the basic classes that point more to the intrinsic notion of writing itself. But not all scholars do so. In the last two decades, a traditional term for arrangement, or abecedary order, became almost totally intermingled with what grammatologists hold important in terms of ‘alphabet’ type—with fatal efffects on terminology. Maybe some would object that the abecedary order is not relevant. But the chosen term suggests that the order of signs is relevant— which, indeed, it is. Sadly, this is how terms got confused. It was Peter T. Daniels who brought up a terminological opposition between Alphabet and Abjad. He fijirst did so in 1990 to refute I.J. Gelb’s earlier statement that West Semitic scripts were not alphabets, but rather syllabaries (Daniels 1990, Gelb 1963, see also Powell 1991: 238–245 and Powell 2009: 153). Meanwhile, this distinction has attained certain notoriety amongst linguists and, above all, on many Internet sites. Respond-

20

 It is clear that the Ugaritic cuneiform alphabetic writing was influenced qua technique (extrinsic factors) by the Hittito-Mesopotamian writing traditions. Nevertheless, it shares its basic intrinsic principle (alphabet) with the bulk of Semitic linear writing. On the other hand, a close investigation of the Northwest Semitic script traditions of the fijirst half of the fijirst millennium BCE has shown that script changes up to seemingly diffferent ‘national scripts’ depend almost exclusively on changes in the scribe’s hand and the scribe’s attitude, that is: on regional sweep and influence of dominant scribal schools (van der Kooij 1986: 90–93.244–251.253). As long as there is no better explanation, we must assume that also the earlier change and fijixing of writing direction(s) had the same cause. Once a dominant writing direction had been set (however marginal the reason might have been), the canon of ‘path dependence’ became the rationale for all further development and did not allow any reconversion of direction or shape in writing.

27–30–22–26 – how many letters needs an alphabet?

23

ing to an earlier objection to Gelb by Swiggers 1984, who diffferentiated between a ‘standing-for’ and a ‘denotation’ view, and because “distinct types of script should not be lumped together” (Daniels 1990: 729), Daniels wanted to further diffferentiate between terms than just ‘syllabary’ and ‘alphabet’ alone. This is why to defijine Alphabet more rigorously he introduces the traditional Arabic numeric letter sequence Abjad as a new script term into grammatology, and in the subsequent years he has vigorously defended it. An alphabet, Daniels claims, “contains characters that denote all or most of the individual segments of a language, both vocalic and consonantal. . . . The West Semitic scripts, then, are not alphabets, since they do not represent the vowels.” Rather, the West Semitic scripts “constitute a third fundamental type of script, the kind that denotes individual consonants only” (Daniels 1990: 729), for which he proposes to use the traditional Arabic numerological term Abjad. Only the Greeks, Daniels claimed, fijirst developed what one could label a ‘true’ alphabet. However, he did not say so because of the etymology of the word (which would be a true and irrefutable reason), but because a ‘true’ alphabet in his view must represent all or nearly all phonemes of a language including vowels—which obviously is not the case with any ancient West Semitic ‘alphabet’. To escape objections regarding the vowelled aleph-signs in the Ugaritic expanded 30–letter Abjad for instance, he simply labels this writing system an ‘augmented Abjad’ (Daniels 1990: 730). By the way, he even introduced a “fourth fundamental type of script,” though not as new, when he along the way redefijines the older term ‘neosyllabary’ for the Ethiopic script system and labels it Abugida (Daniels 1990: 730).21 It is fully comprehensible that in his zeal to refute the Gelb thesis, Daniels creates another and more sophisticated system of his own. Some years later only, in astounding aplomb he then stated: “It must simply be recognized [. . .] that abjads are not (any longer) syllabaries and not (yet) alphabets, and that abugidas—though they denote syllables—are not like syllabaries, since vowels receive identifijication equivalent to that for consonants.” (Daniels 1996a: 8). Notwithstanding the fact that the apparently sophisticated nomenclature used by Daniels has already found wide

21  It is interesting that Daniels here is still cautious: “Were it not for the existing term, I would propose maintaining the pattern by calling this type an ‘abugida,’ from the Ethiopian word for the auxiliary order of consonants in the signary.” However, some years later, the Abugida is another fijixed Daniels’ term passim in The World’s Writing Systems, ed. by Daniels & Bright (1996).

24

reinhard g. lehmann

acceptance in many publications, it also has its awkwardness. In his quest for the early history of the alphabet, such distinctions weaken the eye and obfuscate terminology instead of clarifying it. So the remaining question is: what must simply be recognized? Granted that the distinction between Alphabet and Abgad were to be helpful from a typological and grammatological point of view, however, is it also correct or even useful regarding linguistics and Semitic philology? I am not so certain about this. Regarding Semitic writing, Daniels’ distinction is artifijicial, and it is also dubious regarding Semitic philology and linguistics. Rather, this distinction is again to foster a certain bias as if the West Semitic alphabet were not yet entirely complete, lacking something important to be a fully working script system.22 In rejecting concepts of ‘evolution’ with respect to the history of writing, Daniels opts for “successive improvements” (Daniels 1996a: 8). This though is awkward, too, because not every change in a system is necessarily

22  What is worse (however, Daniels is not to be charged with this), the distinction in favour of a ‘true’ Greek ‘alphabet’ is capable of strange eurocentristic or at least graecocentristic chauvinistic efffects, as Daniels himself pointed out by rebutting Eric Havelock (Daniels 1996b: 27–28). Notions similar to Havelock’s can also still be found in recent German publications, for example where of the Greek alphabet it is said that “die bahnbrechende Entwicklung eines Alphabets mit Konsonanten- und Vokalzeichen strahlte sowohl nach Osten aus [. . .], als auch nach Westen [. . .]” (Casaretto, et al. 2007: 38), and similarly: “Die Suche nach allgemeiner Kenntnis von den allgemeinen Gesetzen der Natur, die insbesondere der Medizin eine völlig neue und bis heute wirksame Fassung gab, war auf leistungsfähige Notationen angewiesen. Den Griechen kam zur Hilfe, dass sie ein gleichermaṣen exaktes wie flexibles Schriftsystem entwickeln konnten. Es förderte die Abstraktion, die man zur Erkenntnis gesetzesmäṣiger Vorgänge in Natur und Gesellschaft benötigte” (Gerhard 2007: 6). It goes without saying that this corresponds neither to basic historic facts nor to the historical deductions therefrom. Probably, the most radical German exponent of such a notion is the media theorist Friedrich Kittler (however, note the recent caveat by Jan Assmann 2011: 110–111). Unfortunately, also the esteemed classicist Barry P. Powell (2009: 153fff) clings to a certain graecocentrism with fatal consequences for his perception of Semitics. Powell is at least poorly informed, when he contrasts “fewer than one hundred examples of West Semitic writing from the Levant” surviving from the whole fijirst millennium BCE with the “epigraphic testimony to the Greek alphabet, swelling into an ocean of epigraphic remains” (2009: 153–154). Obviously, his volitional graecocentrism in matters of writing had let him astray regarding what the current state-of-the-art of West Semitic epigraphy really is (and already was during the past 50 years or more). Actually Powell would had done better at least to ponder on what Tatian noted already in the second century: “Do not maintain a totally hostile attitude to foreigners, men of Greece, nor resent their beliefs. For which of your own practices did not have a foreign origin? The most famous of Telmessians invented divination through dreams, Carians foreknowledge through stars; Phrygians and the most ancient of the Isaurians the lore of bird-flights, Cyprians a cult of sacrifijices; Babylonians astronomy, Persians magic, Egyptians geometry, Phoenicians education through the letters of the alphabet. Therefore stop calling imitations inventions.” (Tatian, Oratio ad Graecos 1.1; Whittaker 1982: 2–3).

27–30–22–26 – how many letters needs an alphabet?

25

an improvement of a former system, which then might be abandoned. Rather, at least in the most ancient Semitic alphabetical systems changes seem to be mere adaptations or adjustments to fijit better the requirements of a given linguistic situation. It remains problematic to take over the term Abjad in its ‘Daniels’ defijinition when dealing with the early history of the (West) Semitic alphabet. Even the late M. O’Connor got into confused terminology—coining terms like “southern linear abjad” (O’Connor 1996a: 98) or “South Semitic Abjad” (which nota bene denotes the Halaḥama) and wrote of “distinguishing the alphabet from the alphabetic order . . . to recognize the Levantine order and the South Semitic order, which must be independent of each other, as being of comparable antiquity” (O’Connor 1996b: 790). Fatally enough, there was no unambiguous ‘Levantine’ order, as the rare but real Ugaritic Halaḥama evidence proves, but only an as yet statistically signifijicant preference of Abjad (Abjad)-order use. Moreover, Daniels himself fails to discuss or even to mention the ‘South Semitic’ Halaḥama device that automatically would contradict him by its sole existence: one is curious about how terms would change if one day it happened that an Abjadsequence were found in Sabaic script within Sabaean boundaries, then perhaps speaking of halḥamic Abjads and abjadic (abgadic) Abjads . . .? To take a defijinition by David Crystal, from a linguistic point of view an alphabet simply is “a type of writing system in which a set of symbols (letters) represents the important sounds (phonemes) of a language” (Crystal 1992: 14)—and one feels inclined to complement this with the second half of Daniel’s defijinition of writing in general: . . . “in such a way that it can be recovered more or less exactly without the intervention of the utterer.” (Daniels 2007: 55). The sole point in question is whether a phonemic-graphemic alphabetic system works economically with its language(s) or group of languages, which it was created for or adapted to. Or, to say it in other words: what are the important sounds (phonemes), or what sounds were considered important by the majority of speakers or scribes of a language to recover an utterance more or less exactly without the intervention of the utterer at the time when a phonemic writing system for this language came into being? The answer is clear: Greek is an Indo-European language with a large number of vowel phonemes and diphthongs and a lot of consonantal clusters, all of which are far more semantically distinctive and sensitive than in the Semitic languages. Thus, adopting the Abjad in the Aegean abets adapting it to the peculiar needs of their language type. By and by, assigning some graphemes of unknown sounds to vowels (some more

26

reinhard g. lehmann

dropped out of use: koppa and digamma), and adding some more at the end, Greek fijinally created the Alpha-Beta, which fijits almost perfectly as an alphabetic base system for the non-Semitic Indo-European languages, and which fijinally made the game. This is not even rebutted by the few IndoEuropean languages that have been or are written with adaptations of Arabic or Hebrew script, because these adaptations include the additional and later vowel denoting tokens of these scripts as well. On the other hand, by no means is the (west) Semitic Abjad a nonalphabetical system, and by no means is it defijicient (Coulmas 2003: 113). It is Daniels himself who concedes that “the Semitic abjads really do fijit the structure of Hebrew, Aramaic, and Arabic very well [. . .] since the spelling ensures that each root looks the same through its plethora of inflections and derivations” (Daniels 1996b: 27)—which is only partly true at all, if one regards such phenomena as assimilation and metathesis or Phoenician Sandhi at least in Northwest Semitic of the fijirst millennium. The long Abjad system, enhanced by two or—depending on how one counts— three occasionally-used vowel signs (matres lectionis) works perfectly until today with the Arabic language, as is proven by the (modern) Arabic script itself (and is indirectly and unintentionally proven even by Daniels 1997), which has remarkably few diffferences from the Ugaritic long alphabet. And even the short Abjad, though with more frequent mater lectionis use, runs with Modern Hebrew. Thus, an Abjad truly is an alphabet in the linguistic meaning of the word. Abjad or Alphabeta—it is mostly a matter of how much ambiguity one can tolerate and how much entropy one can stand before the writing becomes incomprehensible. “Given the systematic nature of [Semitic] consonant writing, it is clearly mistaken to look at it as something incomplete, an imperfection of technology which was to be fully developed only by the Greeks” (Coulmas 1996: 92). Recently, Jan Assmann also objected to connatural graecocentristic positions as those of the German media theorist Friedrich Kittler: Wenn behauptet wird, die Vokalisierung eines konsonantischen Alphabets komme einer Kulturrevolution gleich, halte ich dies für eine Überdramatisierung. (Assmann 2011: 110)

As a set of letters, an alphabet is not the mirror of what should be there in a language from a phonemic or even phonological point of view, rather, it is the data stock of what provides maximum efffijiciency with least efffort from a semantic point of view—which implies: the driver is the principle of least efffort!

27–30–22–26 – how many letters needs an alphabet?

27

Anyway, the quest for the arrangement of letters is by no means irrelevant. Hence, the Abjad term, which as a numerological term originally denotes a certain letter sequence, remains an important terminus technicus exactly in this meaning, i.e., to denote the very letter order beginning with –͗ b–g–(ḫ)–d–h–w . . . and should not be abandoned and ceded to grammatology. As a minimal terminological compromise, and to further avoid terminological confusion—which may be an efffect of inconsiderately taking over Daniels’ terms—I suggest using Abgad (with “g”) to denote the Northwest Semitic alphabetic device (see below) in contrast to Abjad, which unfortunately already seems to be established for a diffferent notion in terms of grammatology. Halaḥama or Abgad? How to Sort an Alphabet As mentioned, the fijirst basic parameter is the arrangement of letters, that is: the alphabetic sequence, Halaḥama or Abgad? In spite of some intrinsic variants and alterations, and in spite of certain exceptions, it is a matter of fact that, beginning with the fijirst millennium BCE, one can ascribe both long abecedaries, the Halaḥama and the long Abgad device, to the two greater areas of their graphemic type, i.e., the South Semitic and the North Semitic graphic letter sets respectively.23 Anyhow, it is noticeable and most interesting that the letter type that predates them both is the Ugaritic wedge alphabet. It is the Ugaritic cuneiform script alone that displays both devices, the Halaḥama and the Abgad, though by far not on an equal footing. Even though obviously the Halaḥama never was the standard in the realm of North Semitic and vanished in the Levant before the fijirst millennium, and acknowledging the as yet unrevealed roots of the South Semitic script, it is almost indisputable that South Semitic Halaḥama and North Semitic Abgad are siblings (Tropper 1994: 298f.300, cp. Lemaire 2008: 50). This becomes clear not only from their early coexistence in the Northern Levant in the second half of the second millennium BCE, but also from their common phoneme inventory.

23  However, note that there seems to be a certain ‘zone of uncertainty’ in the Ancient North Arabian, where letter inventories or abecedaries are found in (modifijied) Abgad, Halaḥama or simple order by graphic shape (Macdonald 1986, and infra n. 8). This so far unparalleled diversity is explicable by the spread of literacy throughout “ ‘non-literate’ nomadic communities which [. . .] thought of and used the ability to write in a quite different way to the sedentaries.” (Macdonald 1986: 115 and fff.)

28

reinhard g. lehmann

For comparison, we arrange the South Arabic according to the (Ugaritic) long Abgad device: Ugaritic

–͗ b–g–ḫ–d–h–w–z–ḥ–ṭ–y–k–š–l–m–ḏ–n–ẓ–s– –͑ p–s̀–q–r–ṯ–ġ–t– i – u –s̀ –͗ b–g–ḫ–d–h–w–z–ḥ–ṭ–y–k–š–l–m–ḏ–n–ẓ–s– –͑ f–ṣ–q–r–ṯ–ġ–t–ś–ḍ South Arabic / Sabaic (arranged in Abgad-mode)

It can easily be seen that there is a diffference of only two or three signs respectively—the secondarily added Ugaritic signs in their normal position at the end, after t, and two more Arabic graphemes (in the above chart these latter ones are not added at the end of the sequence, it seems so only because the ‘Arabic’ inventory is arranged here in ‘Abgadmode’ for better comparison). The diffferences are indeed few, but they show that already in the late second millennium each must have had its own history. Since the graphemic surplus of the Ugaritic 30-letter long Abgad is added at the end (letters i͗ – u͗ –s̀), and because there is a common phonemic inventory of 27 signs in both the Ugaritic and the South Arabic, although diffferent in sequence, and because the South Arabic inventory has two more letters which are unknown in the Abgad tradition, it is most probable that they both represent an expanded superstrate alphabet, the basis of which must have been a stock inventory of 27, either as an Abgad, or as a Halaḥama device. Nevertheless, it remains unclear when the two devices separated from one another (Schippmann 1998: 18–19)—if they where once united. The controversy becomes even harder because it applies to the basic question how the alphabet came into being at all and which of the two devices, the Halaḥama or the Abgad, is the older one. But this is not the topic here. Short or Long? How Many Letters Does an Alphabet Need? The second parameter is the number of graphemes. Short or long? How many letters does an alphabet need? There is as yet no clear-cut archaeological and hard epigraphical data indicating that either the short alphabet or the twenty-seven-sign alphabet preceded the 22-letter standard alphabet of the fijirst millennium BCE (Pardee 2003/04: 18 n. 45, Pardee 2007: 183f). Some scholars assume that the long Alphabet is an expanded version of the older short alphabet of 22 signs (enhancement theory, for instance

27–30–22–26 – how many letters needs an alphabet?

29

Dietrich & Loretz 1988: 141–143, 1991). However, most scholars argue the other way round that the Ugaritic Long Alphabet (30/27) is an adaptation of an early form of the Northwest Semitic long Abgad device, which fijinally was reduced to 22 signs (reduction theory). The aporia is that so far there is no clear and unambiguous Northwest Semitic linear rich-inventory, i.e., long (27/30) Abgad (or any other device) abecedary listed. Notwithstanding W.F. Albright’s earlier claim (Albright 1950: 12–13) that there also must have been a 27-letter-linear long Alphabet, all attempts to unambiguously detect more than the well-known 22-letter linear Abgad have failed.24 But although the majority of the Ugaritic abecedaries is based upon the Abgad system, it is only Ugaritic that displays all types of Semitic alphabets as yet known to us: long (30-enhanced) Abjad, (27) long Halaḥama, and short (22) Abgad. And it is notable that the full inventory alphabets (27/30) seem to be northern Levantine only, while the reduced-inventory texts are found at Ugarit, too, but more often to the south in Lebanon and Palestine, which means that it was scattered throughout the Levant. Thus, to invoke Occam’s razor, it seems most reasonable to adhere to the reduction theory. This posits that near the end of the second millennium in the Northern Levant an old, inventory-rich, alphabetic system of 27 consonant signs, which was almost exactly matching the consonantal inventory of the spoken language before, became reduced (Pardee 2007: 183f). As matters stand, i.e., because there is not a single linearized instance of a long Abgad surplus letter, for instance /ṯ/ or /ḫ/, we must also admit that by linearization of the graphemic type this reduction to 22 graphemes was already done, or in other words: the 22-letter short Abgad predates or is simultaneous with the emergence of linear letter forms, i.e., the linearization. This implies that the reduction of the Abgad to 22 graphemes must have happened not later than in the third quarter of the second millennium, i.e., most probably in the thirteenth century. This is confijirmed by the date of the Ugaritic-written short Abgad texts. As we now know, it was this trimmed down system that fijinally made the game. Its Levantine development and slight only alteration in the fijirst millennium BCE gave birth to the Hebrew, the Aramaic, the Syriac, the Palmyrene, the Nabataean and others, and fijinally to the Arabic alphabet. By its transmission to the Aegean and further west it became subject

24  For instance M.F. Martin, 1962. A Twelfth Century Bronze Palimpsest, Rivista degli Studi Orientali 37: 175–193 (see infra note 18).

30

reinhard g. lehmann

to various alterations, modifijications, and enhancements, but fijinally was spread throughout the world. Shortening the Alphabet I: Who and When? Admittedly, there is not a single primordial vowel sign in the Semitic alphabets. However, there is a need for grammatologists to understand that this is no impediment to understanding what an alphabet is. Also it is not the problem that we have a respectably high amount of letters, including vowel signs, in the Greek Alphabet. And it is also not a challenging question here why there are so many consonant-only letters in the South Arabic or in the Ugaritic writing tradition. The most puzzling question is: What about the short Abgad system? Why are there so few letters left in the Northwest Semitic Abgad of the fijirst millennium BCE? Of course languages with a rich phoneme inventory can get along with a defijicient, or reduced alphabetic grapheme inventory (Coulmas 2003: 113). On the other hand, “a borrowed alphabet in which phonemes of the borrowing languages are lacking tends to become polyphonic,” (Blau 1982: 3, cf. Blau 2010: 73fff)—which is only the reverse of the medal. For example, older Aramaic obviously had at least 26 consonantal phonemes (Creason 2008: 112–114), but is written with the 22-Abgad system, and similarly this is the case with Hebrew for instance. Hence one cannot help admitting that the speakers of Hebrew or Aramaic adopted an incongruous alphabetical system, which beforehand had already been reduced to only 22 letters, insufffijicient to represent all consonantal phonemes.25 There again is no just cause to assume that their common donor or forebear never had more than 22 graphemes. The sole existence of a long Abgad tradition, though in cuneiform only, but nevertheless Abgad, makes a strong point against that. But if a sudden reduction of the Abgad alphabetical system from 27 to 22 graphemes had taken place at some point more than 3000 years ago, where and when did this occur, and why was it done so? Most scholars assume that the (Proto-) Phoenicians did it, and there seems to be no reason to object to it (Lemaire 2008: 49f ).

25  The same, to be sure, happened again more than a thousand years later when Arabic took over the Nabataean-Aramaic alphabet.

27–30–22–26 – how many letters needs an alphabet?

31

To be sure, there is no clear-cut evidence that the Phoenicians invented the alphabet, but there is less evidence that the Hebrews (or others) did. Thus, to invoke Occam’s razor again, it seems most likely that the early Phoenicians were, if not the inventors of the 27-letter alphabet, at least the transmitters of the 22-letter alphabet. The very earliest texts of considerable length known to us that were written with the linear-short Abgad are undoubtedly Phoenician. These come from Byblos, like the famous Ahirom sarcophagus inscription (Lehmann 2005a, 2008b) or the Shipitbaal, Abibaal, and Elibaal inscriptions, and somewhat later also from other Phoenician sites like Tyre, Sidon, and the overseas colonies. Moreover— putting aside the inscribed Canaanite arrowheads of the late second millennium—, even the earliest of these texts already at the turn of the millennium display a considerable degree of scribal experience, skill, and craftsmanship (Rollston 2008a, Lehmann 2008b). What we know of the original West Semitic texts, i.e., inscriptions, from the middle or late second and even of the fijirst millennium, are sherds, grafffijitti, and fragments incised in stone, but naturally nothing that is written flat. Flat writing however, i.e., with ink on papyrus or another smoothed or planar surface, must have gone on simultaneously with and even earlier than clumsy scratches on sherds. This is also most likely considering the conjectural genesis of Ugaritic cuneiform (Lundin 1987b, Dietrich & Loretz 1988). It is also evident that scratched or carved ‘cursive’ linear letter forms in argil always are predated by plain cursive types of professional flat writing (van der Kooij 1986). Unfortunately, all this has gone forever because of the perishable nature of papyrus in the climate of Palestine, and, therefore, conclusions are speculative. The dearth of evidence for the early history of the linear alphabetic writing obscures our knowledge of the origin and early history of the alphabet. Nevertheless it has particular implications. Some years ago, Benjamin Sass emphasized the sudden thirteenth-century alphabetic boom after a purported “excessively long palaeographic standstill” of ca. 1800–1300 BCE (Sass 2004–05: 148). This fijinally causes him to doubt the existence of (non-linear) alphabetic inscriptions preceding the fourteenth century, and to claim that the early history of the alphabet was not so early at all, but that the alphabet was born in Palestine in the thirteenth to twelfth centuries, shortly before it lost its ‘Proto-Sinaitic’ look and letter shapes became linear. However, notwithstanding the problematic dating of some early non-linear alphabetic inscriptions, Sass’ view is loaded with some methodologically problematic and romantic presuppositions, as, for instance, in this passage:

32

reinhard g. lehmann Rather than as a sudden ‘boom’, its timing quite inexplicable, ending a centuries-long freeze, this surge in the 13th–12th centuries may be understood as the swift, enthusiastic implementation of the breakthrough made in Egypt only shortly before. Brilliant inventions take root fast. (Sass 2004–05: 154)

However, by no means every ‘brilliant invention’ takes root as fast as Sass assumes, especially not if it undermines much older stable and established systems, as were the Mesopotamian cuneiform and the Egyptian writing. A ‘freeze’ or standstill in palaeographic development is only then problematic, if one confijidently assumes that the bearers of an early alphabetic invention, and those who handed it down, were already ‘democratic’ in that sense that they shared their knowledge and skill freely with everyone who was interested. But how can we know whether they did so? The meagre contents of the few intelligible ‘Proto-Sinaitic’, or early nonlinear or proto-linear West Semitic inscriptions, including the recently published and highly controversial Qeiyafa ostracon (Misgav, Garfijinkel & Ganor 2009), give no hint in this or in another direction. Hence, it is also possible that alphabetic writing was the arcane knowledge of only a restricted group of people or functionaries, such as some kind of (w)āšipu(m) or early ‫ ְלוִ יִּ ם‬lǝwiyyím (ritual) specialists for instance, or other.26 As long as there was no broad social backing (and the evidence does not point in this direction), a new, phonematic ’bgd or hlḥm based script system, which demands a good deal of abstract efffort from its users, could not have ‘boomed’ suddenly. Alphabetic writing is not such easy a skill to acquire in a short period of time—either by a gifted individual or even by a social group accustomed to writing, if any, in non-alphabetic systems, as often was assumed regarding the development of the Northwest Semitic alphabet. Rather, acquiring any fluent or even orthographically correct profijiciency in writing a fijirst alphabetic system requires substantial time (Rollston 2008b: 68–69). Moreover, special calligraphic knowledge, as it was already employed in the earliest Phoenician inscriptions of Byblos (Lehmann 2008b) but is lacking in the ‘Proto-Sinaitic’ texts, needs a developmental period. This equally rules out an invention of the alphabet shortly before these inscriptions. Hence, one has to conceive a considerable incubation period in which the ‘idea of alphabet’ could steadily grow and stabilize as an efffec-

26  Lemaire (2008: 49) speaks of “priests of local shrines who preserved and developed the use of this ‘Canaanite’ alphabetic script, perhaps so as to express a certain specifijic religious and cultural tradition.” From an ethnographic perspective, cf. also Schulz (1987).

27–30–22–26 – how many letters needs an alphabet?

33

tive and competitive system, before it could achieve currency. “Thus, in the current state of the documentary evidence, the most credible working hypothesis would seem to link the origins of the alphabet . . . around the middle of the period 2000–1300” (Lemaire 2008: 47). No fast-taking root of some ‘brilliant idea’ was responsible for the boom and sudden spread of alphabetic literacy. Rather, we have to look for other ‘booming factors’, which are equally able to explain an astounding—and maybe sudden—reduction of the alphabet in its Abgad device. Shortening the Alphabet II: Who Did What? If it was the (Proto-) Phoenicians who boiled down the alphabet to 22 graphemes, and if their way of doing so immediately ‘boomed’ all over the Levant, why did they do so? We remember what the wise calligraphy master Serani said in Rafijik Schami’s novel: “The Arabic language has only twenty-nine letters, and the more of them you destroy, the more uncertain and imprecise the language becomes.” Would it not be correct to say, that the fewer letters a script system has the more uncertain and imprecise the meaning? Semiticists know about the difffijiculties in deciphering Phoenician inscriptions because the defective and vowelless grapheme system doesn’t even use supportive vowel letters, i.e., matres lectionis. The most prominent and traditional explanation seems to be that widespread linguistic changes had taken place in that period of time, which involved the loss of certain sounds. This is described as a typical feature of the Phoenician language. In grammars and concise descriptions of Phoenician, it is generally assumed that the (spoken) Phoenician language had only 22 phonemes, which implies a full match of the graphemic with the consonantal phonemic inventory, i.e., that Phoenician had as many graphemes as it had consonant phonemes. (Harris 1936: 16f; Swiggers 1991: 119f, Friedrich & Röllig & Amadasi 1999: 11; Krahmalkov 2001: 16.19,27 recently again Krebernik 2007: 124, Rainey 2007, MacDonald 2008b).28 The 27  However, Krahmalkov (2001: 19) notes that in early Phoenician of the Late Bronze Age the consonantal repertory was larger. Thus he self-contradictorily objects to the commonly held opinion that the 22-letter Phoinikeia grammata (φοινικήια γράµµατα or καδµήια γράµµατα, Herodot Hist. V 58–59) alphabet was invented by the Phoenicians but, rather, “had been devised in the Late Bronze Age and later” and was “adapted by the Phoenicians to the needs of their language which, in the early Iron Age, possessed a repertory of twenty-two consonantal phonemes” (Krahmalkov 2001: 16). 28  Maybe one of the most pithy precis of this respect is indeed made by MacDonald (2008b), who states that “it was singularly unfortunate that the fijirst widely used linear

34

reinhard g. lehmann

assumption is drawn from the fact that the Phoenician alphabet—which is claimed to be the oldest standardized linear alphabet in the Eastern Mediterranean—has only 22 graphemes. And normally there are no further questions or doubts on that. Most notably, already in 1936 Zellig Harris, who was one of the most influential scholars in the twentieth century regarding Phoenician phonemics (on Harris, see Barsky 2011), stated that “the complete absence of spelling variations in the Phoenician inscriptions as far back as they go would indicate that the alphabet conformed exactly to the needs of the language.” However, Harris clearly saw and expressed the methodological weakness of the argument, admittedly also stating in an often overlooked footnote that “this is, of course, merely an argument from silence.” (Harris 1936: 17 n. 21)29 The assumption equally implies that the Hebrews, the Aramaeans, and the other people of fijirst millennium Levant speaking a non-Phoenician Northwest Semitic dialect, or language, unhesitatingly and invariably followed the Phoenician example in the trail of a supposed alphabetic ‘boom’ (Sass 2004–05) of the late second millennium BCE to content themselves with 22 graphemes only. But why should they have done so? Alan Millard recently again pointed out that, “as ancient people read aloud, it may be assumed the signs gave sufffijicient information . . .” (Millard 2007: 85). Texts were recovered through re-oralization, which means that the message was re-activated only by enunciating again that which was written down, thus, by reconversion of visible script into audible speech. Compare for instance the semantic development of the verb qr ͗ ‘to shout’ > ‘to read out loud’ > ‘to read’. If this is true, the most intriguing point seems to be how the reconversion process from visible script to audible speech could have worked in an area, where a variety of presumably diffferent Canaanite dialects with diffferent phonemic treatments of the interdentals and the laryngeals still existed. It is widely accepted that the main Canaanite dialects including

alphabet was designed to express Phoenician, which had one of the smallest repertoires of consonantal phonemes of any Semitic language. Alas, the twenty-two letters of the Phoenician alphabet were treated as sacrosanct within the Near East, and the non-Phoenician Near Eastern languages which came to be written in it were squeezed into this rigid frame, regardless of the resulting ambiguities.” 29  It should be also noted that already Harris made only one page later a possibly unconscious admission to a non-congruent phoneme-grapheme-relation in Phoenician writing: “The spelling in these inscriptions is etymological and the rules of orthography are rigid and unchanging, hiding all changes in pronunciation.” (Harris 1936: 18, italics by R.G.L.).

27–30–22–26 – how many letters needs an alphabet?

35

Hebrew, including also the Transjordan fringe, and also the Aramaic varieties of the fijirst millennium, despite of their 22-graphemes-only script system employed more than consonantal 22 phonemes in their speech at least up to the seventh or to the second century BCE respectively. The proof was already made by Brønno 1970, Wevers 1970, Blau 1982, Degen 1969: 32fff, and Garr 1985, though with diffferent methods, and many scholars had contributed to this topic in the last decades with further observations and adding more comparative material to subsequently refijine the picture (Sivan & Cochavi-Rainey 1992; Hoch 1994, Steiner 2005). Today, it is generally accepted that in the fijirst millennium BCE for the vast amount of Northwest Semitic languages the 22-letter alphabet was defijicient from a phonemic point of view, or, in other words: the data reveal that at least to some extend sounds like the interdentals /ḏ/, /ṯ/ and /ṯ/̣ had not yet merged to their neighbouring sounds, and that often also the velar-pharyngal pairs of /ḫ/—/ḥ/ and /ġ/—/ /͑ were still phonemically distinct and retained in speech at least in the late Bronze and Early Iron Ages—and further on. It is true that Phoenician orthography remained conservative, rarely indicating vowels even after routine contact with the diffferent devices to denote vowels in Greek, or, by use of matres lectionis, in Aramaic and Hebrew. But what, on the other hand, are the reasons to assume that the Phoenicians, only some hundred kilometres south from Ugarit with its rich phoneme and grapheme inventory, should have had a thus remarkably decreased consonantal phoneme inventory, dropping the interdentals and other sounds? Claiming that the Phoenician alphabet employed only 22 letters because that number represented the inventory of consonantal phonemes in their language means alleging that the 22-letter Abgad was a real phonemic ‘alphabet’, or system, at least for Phoenician. But is this really true? Since consonants carry the gist of information in Semitic languages, is it really feasible to say that the Abgad, once it lost some of its graphemes like ‹ḫ›, ‹ġ›, ‹ḏ›, and ‹ṯ›, was still purely phonemic? Or isn’t that turning an outcome into its cause? Granted also that there are no Phoenician texts in a 27-letter alphabet, but only in a 22-letter system: is this sufffijicient reason to argue that they introduced the 22-letter short Abgad because they did not have more sounds in their language, whereas their eastern and southern Aramaic, Israelite, Judahite, and Moabite (recently Beyer 2010: 10–17) neighbours continued to articulate discernibly somewhat more than 22 phonemes in their dialect or language? Moreover, is it really within reason to imagine that those speakers simply jumped on the bandwagon of such mumbling

36

reinhard g. lehmann

22-sounds-only-writing Phoenicians?30 From a methodological point of view there seems to be a considerable circular reasoning in these assumptions: it jumps too hastily to an explication, which anticipates the outcome (Phoenician script bequeaths only 22 graphemes) to be the cause (Phoenician speech had only 22 consonantal phonemes). Even those who freely admit that there is inevitable material demonstrating that by no means was the early Phoenician language of the Levant a consonant-poor language and the 22-letter short Abgad simply displayed its actual phonemic state, do not try to escape the traditional view that the Phoenicians were responsible for the truncation of a formerly longer alphabet. Rather, one tries to postulate the short alphabet as the reflex of an actual “innovating speech of the royal elite”, thus introducing a ‘short speech’ as a kind of sociolect that fijinally “would also be used for the speech of the common citizens” (Hoch 1994: 415). However, this is a way to subsequently reduce the derivation group of the short Abgad by and by into nothing: which royal elite where? Anyhow, a last retreat from which to argue the point of a 22-letter oneto-one grapheme-phoneme correlation again and again still seems to be Phoenician, which is held responsible for the spread of the short Abgad even over an area where languages with more than only 22 phonemes still were spoken. Nevertheless, these languages continued for quite some time, possibly for the duration of their existence, to write and spell in accordance with a presumed phoneme-based Phoenician—hence for themselves defijicient—orthography. Shortened Phoneme Inventory: How Many Phonemes? If it is true that Phoenician and the Phoenicians were responsible for the spread of the Short Abgad system (and the epigraphic evidence from Byblos for instance still points in that direction), there is another difffijiculty that has come to the attention of scholars in the last decades: is it

30  Accordingly, Rainey (2007: 69) from an entirely diffferent approach and with a view to ‘redefijine’ Hebrew as a Transjordanian language simply claiming that “the speakers of Hebrew did not speak the same dialect as those from whom they borrowed the alphabet”, recently states that these borrowed the Canaanite (= Phoenician) alphabet because “the rustic clans from the steppe lands were so impressed by that superior cultural feature that [. . .] they adopted the writing medium of the highly cultured people of the coastal areas.” Rainey correctly points out the question of an obviously defijicient graphemic system for these languages. However, his explanation seems to be as romanticized as that of Sass.

27–30–22–26 – how many letters needs an alphabet?

37

still possible to argue that Phoenician had 22 phonemes solely because the ‘Phoenician’ alphabet had only 22 graphemes? This became one of the key assumptions for the reconstruction of Canaanite. However, the underlying syllogism is not compelling, and by no means is it mandatory or self-evident that early Phoenician should have had 22 consonant phonemes only. Without any claim to be comprehensive, let us have a look on some of the relevant data, mainly from Egyptian sources. The transcriptions of Semitic words in Egyptian writing of the New Kingdom and the Third Intermediate Period as discussed by Hoch 1994 probably tell us more about old Phoenician and Hebrew than current hypothetical reconstructions of these languages can tell us about the origin of the words here studied. The generally accepted reconstructions rest on a very bold assumption—that the short Semitic alphabet contained a grapheme corresponding to each phoneme. It will be suggested that the assumption that the bulk of Semitic words are from older dialects of Phoenician and Hebrew is more likely than the assumption upon which Phoenician is reconstructed. The conclusion to be drawn is that our current reconstruction should be reconsidered. (Hoch 1994: 12)31

From Hoch’s data it follows that “the most reasonable conclusion seems to be that most of the Canaanite dialects had a fairly extensive phonemic inventory and used a short alphabet, with polyvalence of several graphemes”. His data confijirm the polyvalence of certain short Abgad letters in the Late Bronze and Early Iron Ages as employed by virtually all Canaanite dialects, and even so that “Phoenician may have also been more conservative in its phonemic inventory than universally believed” (Hoch 1994: 416.417).32 However, because the source language of Semitic words in Egyptian writing as discussed by Hoch 1994 remains unsure in many cases, only the few well-defijined unequivocal Phoenician examples are listed here.33 Thus, from a methodological point of view, two conditions are to be applied 31  Compare also, towards the end of Hoch’s book: “By far the most important factor that emerges from the Egyptian evidence is the size of the phonemic inventory, which numbers as high as 27–29 phonemes, even in the more recent material. This is far more than usually believed to be present in the contemporary Canaanite dialects. Although it cannot be demonstrated that any single dialect contained the full inventory, it would seem likely that dialects with mergers were in the minority as source languages” (Hoch 1994: 413). 32  Cp. also Krahmalkov (2001: 19), however, see infra note 27. 33  More examples, though not unequivocally classifijiable as proper Phoenician, are found in Helck (1971: 507–527); Sivan & Cochavi-Rainey (1992); Schneider (1992); and Hoch (1994) passim.

38

reinhard g. lehmann

here to Hoch’s material, which causes a substantial restraint of evidence: a) the Semitic foreign word must comprise diagnostic consonants to prove a polyvalence of the Short Abgad at all, and b) the Semitic foreign word must be identifijiable as unequivocally Phoenician by external evidence, thus being a Phoenician toponym or anthroponym, or by the shift of short accented */a/ > /ō/.34 Anyhow, even if every word which is not well-defijined Phoenician by these conditions is ruled out, there are still enough clear-cut arguments left over: In the Egyptian story of Wenamun (1,29; 3,7 and 1,16–17; Schipper 2005: 177), the name of the Phoenician-Byblian ruler Zakarba l͑ , or Zakurba l͑ , is written Ṯ-k-r-b- -͑ l representing the voiced interdental phoneme /ḏ/, clearly distinguishing it from /z/ (Schneider 1992: 256f; Hoch 1994: 372–373, 400 n. 5, 417). This retention of /ḏ/ in the twelfth or eleventh century gives witness that the Phoenicians still retained a phonemic difference between /ḏ/ and /z/ at a time when the short Abgad already was in use.35 Nevertheless, on at least three almost contemporary or slightly later Phoenician arrowheads36 the orthography of this name has ‹z›, which was the only possible orthography in a Canaanite-based 22-letter Abgad system, as is also displayed in somewhat later cognate Hebrew names like ‫ זְ ַכ ְריָ ה‬Zǝkaryā, or ‫ זְ ַכ ְריָ הוּ‬Zǝkaryāhū, and the Aramaic hypocoristicon ‫ זַ כּוּר‬Zakkūr. Because the root *ḏkr (‘to remember’) not only in Phoenician, but also in some earlier Aramaic texts is written zkr (in contrast to later Aramaic dkr), it is most probable that at least until the early fijirst millennium even in Phoenician Canaanite the overall phonemic polyvalence of the short Abgad ‹z›-letter was retained.37 Also the original interdental */ṯ/̣ seems to have been still phonemically diffferent from /ṣ/, as Egyptian writings of the roots r͑ ṣ (* r͑ ṯ)̣ ‘to terrify’, rṣ ( Ḥ and *Ġ > )͑ and Greek Translations (2 Esdras and Judith). Journal of Biblical Literature 124: 229–267. Swiggers, Pierre. 1984. On the Nature of West-Semitic Writing Systems: Aula Orientalis 2: 149–151. ——. 1991. Linguistic Considerations on Phoenician Orthography. In Cl. Baurain, C. Bonnet, V. Krings (eds.), Phoinikeia Grammata. Lire et écrire en Méditerranée. Actes du Colloque de Liège, 15–18 novembre 1989, pp. 115–132. Namur. Tappy, Ron E. & P. Kyle McCarter. (eds.) 2008. Literate Culture and Tenth-Century Canaan: The Tel Zayit Abecedary in Context. Winona Lake: Eisenbrauns. Tappy, Ron E. & P. Kyle McCarter & Marilyn J. Lundberg & Bruce Zuckerman. 2006. An Abecedary of the Mid-Tenth Century B.C.E. from the Judaean Shephela. BASOR 344: 5–46. Tropper, Josef. 1994. Die nordwestsemitischen Schriften. In Hartmut Günther & Otto Ludwig (eds.), Schrift und Schriftlichkeit. Writing and Its Use. Handbücher zur Sprachund Kommunikationswissenschaft 10.1, pp. 297–306. Berlin, New York. ——. 2000. Ugaritische Grammatik. Alter Orient und Altes Testament 273. Münster: UgaritVerlag. Ullendorfff, Edward. 1977. Is Biblical Hebrew a Language? In Edward Ullendorfff, Is Biblical Hebrew a Language? Studies in Semitic Languages and Civilizations, pp. 3–17. Wiesbaden. Vennemann, Theo. 2006. Germanische Runen und phönizisches Alphabet. Sprachwissenschaft 34: 367–429. ——. 2009. Zur Reihung der Runen im älteren Fuϸark. Analecta Septentrionalia—RGA-E 65: 832–861. Vita, Juan-Pablo. 2003. La protohistoria de la lengua fenicia. In A. González Blanco, J.P. Vita, J.Á. Zamora (eds.), De la Tablilla a la Inteligencia Artifijicial I, pp. 395–410. Zaragoza. Wevers, J.W. 1970. Ḥeth in Classical Hebrew. In J.W. Wevers, D.B.Redford (eds.), Essays on the Ancient Semitic World, pp. 101–112. Toronto. Walde, A. & J.B. Hofmann. 1938. Lateinisches Etymologisches Wörterbuch. 3., neubearbeitete Auflage. Heidelberg. Wachter, Rudolf. 1989. Zur Vorgeschichte des griechischen Alphabets. Kadmos 28: 19–78. ——. 1996. Alphabet. In Der neue Pauly. Enzyklopädie der Antike, pp. 536–548. Stuttgart: Metzler. Whittaker, Molly. 1982. Tatian Oratio ad Graecos and fragments. Edited and translated. Oxford: Clarendon Press. Young, Ian. 1993. Diversity in Pre-Exilic Hebrew (FAT 5). Tübingen: Mohr.

NUBIAN GRAFFITI MESSAGES AND THE HISTORY OF WRITING IN THE SUDANESE NILE BASIN Alex de Voogt & Hans-Jörg Döhla History of Writing in the Sudanese Nile Basin The Sudanese Nile basin has a long history of writing part of which concerns a group of Nilo-Saharan languages, classifijied by Claude Rilly (2004) as the North Eastern Sudanic branch. The earliest record of these languages is found in Egyptian hieroglyphic writing in which, as early as the Kerma period around 2500 BC, Meroitic names can be identifijied for local rulers of the Kerma state. More signifijicant is the development of an independent Meroitic script that developed ca. 300 BC and continued no later than the fijifth century AD. This writing system was deciphered by Grifffijith (1911) but the language it represents cannot yet be fully understood. A few centuries after the disappearance of Meroitic writing, the Nubians in Sudan started using a language and script known as Old Nubian. This script is largely based on the Coptic script and its texts have religious Christian content. Although not every aspect of the language is understood, Old Nubian is clearly a close neighbour to both the Meroitic language and some of the later Nubian languages spoken in the region. Between the end of the Meroitic script in the fijifth century and the start of Old Nubian writing in the eighth or ninth century, there were other scripts in the region but they were not designed for a Nubian language. The Meroitic script existed next to Greek texts and Egyptian (Ptolemaic) writing, the latter either in Hieroglyphic, Demotic or Hieratic form. Coptic came about in the second century in Egypt and gained more prominence around the time of the Arab conquest in the seventh century. Latin and Latin script were also present although few Romans ever lived in Nubia. In other words, both Meroitic and Old Nubian were developed at a time that many scripts of diffferent languages appeared in the region. Egypt and parts of Sudan were conquered by the Arabs in the seventh century and this eventually made the Arabic script replace most if not all other scripts in the region by the tenth century. Later conquests by Ottomans added another language written in Arabic script. Only British rule in the nineteenth century saw the introduction again of a Latin alphabet

54

alex de voogt & hans-jörg döhla

and for the instruction of English in Sudanese schools, this continues until the present. Today’s languages that are still related to previous written Nubian languages have in some cases a documented grammar and phonology for which an alphabet was developed by linguists but not necessarily introduced to the speakers of the language. On the contrary, the Arabization of northern Sudan has introduced the Arabic language at the expense of the local languages. Even though these languages are not taught, they continue to be spoken. Individual attempts to develop a script for these languages have been frequent but remained largely unsuccessful and literacy is limited to Arabic (Hashim 2004). The Development of Writing Meroitic and Old Nubian scripts were developed in the context of several other writing systems and they borrowed both signs, values of signs and to some extent the system of signs from neighbouring scripts. Meroitic has clear examples of Demotic in its cursive script and Egyptian Hieroglyphics in its Hieroglyphic variation. Old Nubian has borrowed mostly from Coptic and Greek. Neither in the case of Meroitic nor in the case of Old Nubian was an existing script applied to a Nubian language but, instead, a new script was developed. The application of an existing script to another language, which is particularly common with alphabetization projects in Africa, uses signs and sign values from one system only and adapts the system to accommodate specifijic phonemes. Rarely are new signs necessary, no matter how unusual the sounds. In most cases it is possible to adapt an existing script in such a way that it can represent another language. It is even known that Chinese, Meso-American and Cuneiform scripts have applied their complex systems to other and unrelated languages. In the light of the above, today’s speakers of Nubian languages in the Sudanese Nile valley have at least four options. They can remain illiterate and write Arabic when writing is required. They can develop a writing system of their own, which has happened at least twice before with Old Nubian and Meroitic. They can adapt a Roman alphabet developed by linguists and other individuals. Or, they can apply the regional Arabic script to Nubian. Since there is no offfijicial encouragement to even speak Nubian, the way in which to write their language is left to the speakers themselves.

nubian graffiti messages and the history of writing

55

The Purpose of Writing The moment of invention, the standardization and thereby the introduction of a writing system are dependent on the immediate use of the script. Scripts that enter school and administrative systems are immediately formalized into a ‘standard’ but writing that is only used in informal settings, such as in letters to family or in markings on private property, commonly requires less formal orthographic rules. The Nubians on Sai Island have started to write their language using Arabic script. The government, including the schools and administration, does not facilitate this process in any way. On the contrary, the teaching of Nubian is not allowed within the Sudanese school system. Also, it is not all Nubians in northern Sudan who have started to write. The examples of Sai Island are, therefore, of particular interest, although a complete survey of areas in which Nubian is written is not part of this study, the efffort to write Nubian using Arabic script seems to be a local afffair. The instances in which Nubian is found written in Arabic script mainly concern grafffijiti on carts and other vehicles with messages that relate to the owners. Grafffijiti writing refers here to any type of public marking that may appear in the forms of a single word or an elaborate text. They are unofffijicial but public markings. Nubian speakers on Sai are, therefore, a rare example of a group of speakers that adapted an existing writing system to their own language for informal use and without government or linguistic support. In other words, in this case there is no clear inventor or teacher of the writing system whom we can immediately identify. Nubian in Arabic Script The Nubian and Arabic phonology are diffferent and this diffference concerns mainly a number of consonants. Other aspects of the language such as the number of vowels, vowel length and stress are, at fijirst glance, similar to Arabic. Since vocalized Arabic script allows most Nubian vowels to be written, a writing system based on a linguistic analysis of the two phonological systems is relatively simple for three reasons: – there are few distinctive phonemes in Nubian that are not represented in the Arabic script – there are signs in Arabic script that are superfluous for Nubian and that can be used for unique Nubian phonemes

56

alex de voogt & hans-jörg döhla Aswan

Egypt

Ni

le

Riv

er

Sudan

Sai Island

Abri

Khartoum

Sudan and Sai Island

– linguistic features such as tone that are particularly cumbersome to represent in Arabic and most other scripts, are absent or of minor importance While the above comparison of phonologies suggests a certain outcome of the Nubian way of writing their language, it is known from the development of scripts elsewhere in the world that this linguistic approach is rarely used when there is a local development of writing. Homonymy and polysemy are not only common in the world’s writing systems, it is unusual if they are not present in a script. In the case of modern Nubian, the use of the Arabic script is a recent phenomenon and the examples presented here only show what an attempt by a group of people to write Nubian in Arabic script may look like.

nubian graffiti messages and the history of writing

57

The Samples The collection of writing samples was started in February 2010. With the help of interpreters and villagers a fijirst selection was made and the texts were transcribed and analyzed with speakers of Nubian. Writing grafffijiti is diffferent from writing a long text such as a letter. The understanding of grafffijiti puts diffferent demands on the writing system, i.e., it should be readily understood by using slogans or words that are easily recognized. In contrast, long texts require more consistency since a larger vocabulary is used and the solutions for writing sounds or certain words become more confusing if they are not systematic. No tradition of personal correspondence or letter writing was found on Sai island or its neighboring town Abri but public messages in Arabic script are frequent using both the Arabic and Nubian language. The grafffijiti on the island is particularly prominent due to the government’s plan for a dam in the Nile. Endless painted slogans decorate the walls on the island reiterating “No to the dam” in Arabic. Occasional Nubian texts accompany the Arabic ones. In addition, Nubian and Arabic texts are found decorating donkey carts and other vehicles. Most grafffijiti is found on the side of converted pick-ups for public transport, the so-called ‘buksi’, whose texts have been commissioned by the owner. Their preferred proverbs and messages are found next to messages in Arabic; both are found near the brand name of the car or truck, which is never Arabic. For example, Japanese names in Roman letters are next to Arabic messages in Arabic script as well as Nubian proverbs in Arabic script. The messages are diverse in content. Some are short but personalized and read ‘Ali’s son’, ‘Kurdi’s son’, or simply ‘my life’ or ‘you are good’. Specifijic romantic texts include ‘To meet you is my hope’, ‘I am with you until death’, ‘Why is your heart not kind to me’ and ‘The beautiful one is you’. This last expression is part of a well-known Nubian song. Another text that reads simply ‘Nuba village’ also refers to a song about the Nubian land. This content gives reason to writing texts in Nubian rather than Arabic. In contrast there are religious texts that suggest the opposite and make Nubian a curious choice. For instance, ‘We got our God’ or ‘Look at this but pray for the Prophet’ or, more appropriate for a vehicle, ‘Pray for the Prophet and say a blessing for the car’. Expressions such as ‘Go in peace’, ‘Go and return well’ or ‘If you do not fly, you cannot catch me’ are particularly appropriate. Certain expressions such as ‘To earn money, he is happy’ which means ‘He who worked hard to obtain money, he should be happy with his money’ are specifijically Nubian. These

58

alex de voogt & hans-jörg döhla

idiomatic expressions are more elaborate to place on a pick-up truck. Another one translates ‘Treat the bad speech with the good speech’ or ‘Say goods things to counter those who say bad things’, again a typically Nubian phrase with presumably no immediate Arabic equivalent. None of the phrases that were collected were other than decorative. There were no instructions or other functional messages present that would make the text ‘required reading’. For both Arabic and Nubian speakers, it is not necessary to understand any of the Nubian texts for any practical reason other than to appreciate the creativity of the owner. For the owner they are a personal expression of being Nubian. The Graffiti Script Nubian written in Arabic letters on Sai Island and its neighboring town Abri is not always vocalized and has few conventions that would make it stand out visually from other Arabic texts. In practice, this means that recognizing a Nubian text requires the observer to read the text fijirst and only when it does not make sense in Arabic, to try and see if it is Nubian instead. This process was witnessed repeatedly when searching for Nubian messages. This is partly due to the nature of Arabic writing, the local Arabic dialect and the inconsistencies that are common and have made Nubian written in this way a difffijicult exercise for anyone not fluent in Nubian. Only with longer phrases were Nubian speakers themselves sometimes confused about the correct reading. Arabic letters commonly have an initial, medial and fijinal variation when written in a word. Some letters such as ‘r’, ‘d’ or ‘w’ do not have a medial or initial shape and are always written separate from a letter that follows. This confuses word separation, particularly if another language is being written, although the Nubian readers did not seem to have particular problems with word separation. The Nubian texts use Arabic that is often not vocalized. Only long vowels and end-vowels are frequently indicated. This is diffferent from how Arabic speakers write loanwords, such as brand names or foreign names in advertisements, since they are commonly presented phonetically with vocalized consonants. The Arabic dialect spoken on Sai Island pronounces the ‘q’ (qaf ) often as a velar consonant and this letter is often employed to write ‘k’ or ‘g’. The other consonants follow Arabic consonants with a few exceptions.

nubian graffiti messages and the history of writing

59

Nubian has a palatal and velar nasal that are not present in the Arabic language or script. The palatal nasal is consistently written with the Arabic letter used for a palatal fricative and specifijically designed for loanwords. Velar nasals are not as common as the palatal one in this corpus and was commonly represented by an ‘n’ followed by a velar stop so that a ‘ng’ would sufffijice as a spelling. Some consonants were omitted or written with a diffferent letter but the examples are not sufffijicient in number to conclude whether this is a particular interpretation of the Nubian pronunciation, a spelling error or a spelling variation. A corpus of grafffijiti only allows a fijirst illustration of the writing practice. The amount of consistency points at the way that speakers of a language analyze their language prior to writing. The following three examples illustrate the complexity of reading Nubian with Arabic script: 1. Arabic fā kid mw

Nubian /fa:kidimu/

English ‘we will not sink’

This example is the only one found on a wall rather than a vehicle. It was found next to many Arabic texts stating ‘No dams!’ (see photo 1). It illustrates that diffferent vowels can be represented but that this is not done consistently. This example also illustrates the confounding problem of the Arabic spacing between letters. For instance, there is no medial or initial ‘d’ or ‘w’ so that they are always found as a fijinal letter even if the word itself is still continuing. 2. Arabic ‘lyn br kh

Nubian /ilin baraka/

English ‘you are good’

The absence of the medial or initial ‘r’ and ‘a’ visually splits this phrase in four parts, but an experienced reader of Arabic would distinguish two words. Baraka is a loanword from Arabic so that the Arabic spelling could be used. Note that ‘alif’ is used to represent an initial vowel or glottal stop, the reader is left to fijind out which one. (See photo 2.) 3. Arabic ‘u w skh bnchmyn maskh bnchy maslq ‘w s f ’ qw r d nqn’ ny

This particularly long phrase features a letter that is commonly used for loanwords only. It is a voiceless postalveolar afffricate, presented here as ‘ch’.

Photo 1. Grafijitti on the wall in Nubian (translation in main text).

60 alex de voogt & hans-jörg döhla

nubian graffiti messages and the history of writing

Photo 2. Donkey cart with Nubian text (translation in main text).

61

62

alex de voogt & hans-jörg döhla Nubian /uska baɲimun maskabaɲ maslug us fagurdaŋgenani/

The Nubian phrase shows both the palatal and velar nasal. The fijirst is represented by the ‘ch’ in Arabic and the second usually by a combination of ‘n’ and ‘q’ but in this example the ‘q’ is also pronounced so that only the ‘n’ indicates the velar nasal. The necessary spacing between letters in Arabic, makes this phrase particularly hard to read. The third Nubian word maskabaɲ is clearly two words for an Arabic reader since the ‘h’ that is used as a vowel also has a medial form that was not used. Also, the vowels are not always written and if they are written, they are not consistently written in the same way. Consistency, however, is not a necessary attribute for writing systems to be functional (de Voogt & Finkel 2010). It only makes this writing particularly difffijicult to read for non-native speakers of the Nubian language. (See photo 3, 4 and 5.) Literal translation Do not try bad to speak good speech, good speech bad speech you will forget it.

Approximate translation: ‘Treat the bad speech by the good one,’ or ‘Say goods things to counter those who say bad things.’

The examples show diffferent Arabic ductus. The grafffijiti on the house shows a Sudanese variety of Kufijic script while the donkey cart has a mixture of Naskhi and Kufijic. Naskhi is presented in nice calligraphy on the buksi although the text ran out of space towards the end. It shows the skill of Arabic calligraphy on the part of the maker. Conclusions Nubians on and near Sai Island write their own language despite the absence of an offfijicial or an agreed upon orthography. They use Arabic script that is not vocalized and has few adaptations in terms of changed sound values or new signs. As such, a Nubian text is not immediately recognizable through the presence of many vowels or unusual characters but commonly has to be read fijirst before someone can understand that it is not Arabic. The vowels /a/, /i/ and /u/ for which solutions exist in Arabic are regularly employed. Palatal nasals are rendered by way of a less

Photo 3. Buksi with Nubian texts on all sides of the pick-up truck. Parked near a fuel station on Sai Island.

nubian graffiti messages and the history of writing 63

64

alex de voogt & hans-jörg döhla

Photo 4. First part of the text with orange background.

Photo 5. Second part of the text in blue background. A short Arabic text with green background is visible too.

nubian graffiti messages and the history of writing

65

common letter used in Arabic script that is mainly used for loanwords. The velar nasal is presented by way of an ‘n’ commonly followed by a letter for a velar stop. The influence of Sudanese rather than classic Arabic is found in the use of ‘q’ for the voiced velar plosive. The messages contained in Nubian script may reiterate or confijirm Arabic expressions that are known for the same context. Messages that refer to Islam and have Arabic loanwords are also found. Other expressions were found that were particular for Nubian and may partly explain the necessity for writing Nubian in Arabic script. Writing Nubian in Arabic script does not seem to be a conscious efffort to revive the language and it is only to a limited extent an attempt to convey what cannot be said in Arabic. Rather it is a diversifijication of grafffijiti that allows new expressions that identify the owner of the text as Nubian. Since most texts were found on vehicles it is this ownership that has given written Nubian a function and only in the context of public display. Indeed, other examples in the study of writing systems show that public signs frequently adopt scripts to languages for which they were not designed (de Voogt 2010). It also explains why school teachers and randomly questioned Nubians are not much aware of written Nubian since it appears as a personal matter, although it is present in a public context, rather than a socio-political efffort to be literate in Nubian. The speakers of many languages in Sudan and elsewhere for which no writing system has been developed have rarely applied a script that is known to them and used it for another, unrelated, language. On the few occasions in which scripts were developed for an otherwise unwritten language, the users applied a foreign script only partially and in nearly all cases the application was done by a singular person, sometimes on behalf of an institution, who subsequently taught his or her orthography to others. The Nubian case illustrates that speakers of a language as a group may apply a script to their own language for the use of public signs. The idea of writing might indeed be enough to inspire such an efffort. The history of writing along the Nubian Nile suggests that diffferent scenarios for the development of writing exist in the same region and even for the same language group and that the context of writing speaks more about its development than its suitability to represent the language. Acknowledgements This research was made possible with the generous support of the American Museum of Natural History. I owe particular thanks to Abdulfatah

Photo 6. Multi-lingual sign posted by the Greek-Norwegian mission on Sai Island using Arabic and Old Nubian scripts for the Nubian language.

66 alex de voogt & hans-jörg döhla

nubian graffiti messages and the history of writing

67

Mohammed Ali Kurdi, Abdullah and Osama, as well as Sami Mohammed Elamin, Claude Rilly, Alexandros and Henriette Tsakos, and Vincent Francigny for their help and support. References Grifffijith, Francis L.L. 1911. Karanog: The Meroitic Inscriptions of Shablul and Karanog. Philadelphia: University Museum. Muḥammad Jalāl Aḥmad Hāshim. 2004. Competing Orthographies for Writing Nobíin Nubian. Occasional Papers in the Study of Sudanese Languages 9:215–249. Rilly, C. 2004. The linguistic position of Meroitic. Arkamani, Sudan Electronic Journal of Archaeology and Anthropology. http://www.arkamani.org/arkamani-library/meroitic/ rilly.htm de Voogt, Alex and Irving Finkel (eds). 2010. The idea of writing: play and complexity. Leiden: Brill.

ABOUT “SHORT” NAMES OF LETTERS Konstantin Pozdniakov Socrates: A very simple matter. I may illustrate my meaning by the names of letters, which, you know, are not the same as the letters themselves with the exception of the four, ε, υ, ο, ω; the names of the rest, whether vowels or consonants, are made up of other letters which we add to them; but so long as we introduce the meaning, and there can be no mistake, the name of the letter is quite correct. Take, for example, the letter “beta”—the addition of e, t, a, gives no offfence, and does not prevent the whole name from having the value which the legislator intended—so well did he know how to give the letters names. Hermogenes: I believe you are right. (Plato)

Where do short names of letters come from? Why do we call a letter [ve] in Russian, [vii] in English, [fau] in German, and [uvə] in Spanish? After all, we do know that all these letters are borrowed from the Latin and Greek traditions, which are, in turn, related by succession. Why would a Russian sigh and say, “yo-ke-le-me-ne” 1 and not, for instance, “*yo-ka-el-em-en”, which would match the “correct” short names of letters? Why do we sometimes say Ka-eL-eM flights and never *Ke-LeMe or *Ka-La-Ma? What makes letter names “correct”? How did it happen that we name the letters [ka], but [de], [em], although schoolchildren, before they are taught the rules, persistently tend to universalize vocalization of such names, each time choosing one and the same vowel [pe], [re], [se], [te] and not, let us say, *[po], *[ro], *[so], *[to]? Many generations of pupils, instinctively and individually, have been making the same “mistake” in naming letters, and many generations of teachers have been correcting these mistakes. Still, as it turned out, neither native speakers nor professional linguists have clear answers to the question—where did this norm come from?

1  An euphemism for a Russian curse designated by fijive letters; roughly comparable with “EKLMN”.

70

konstantin pozdniakov

Letters with “conventional” names are easier to explain. Their algorithm of development is well-known: a frequently used (or “important”) word starting with the sound designated by the letter is selected, like Russian 3 [z]—z’eml’a ‘ground, soil’. Quite often, these names appear to be iconic signs where the denotation of this word is depicted symbolically (Greek delta / Phoenician daleth ‘door’). The situation is more complex when it comes to short—spelling— names. The iconic nature of a sign usually disappears, and two factors determining the name of a letter remain: 1. Certain “heredity” exists—when a short name of a letter is taking shape, the name that existed in the donor language is often preserved; 2. There is some agreement, a “convention” that might not be related to linguistics at all but meets the ideas of a certain group of people responsible for written language and believing that a letter should be named this way and not otherwise. The history of short names is the history of multiple conventions (often forgotten) that were based partly upon pronunciation of letters in the source language and partly upon individual attempts to adjust the letters of the source language and their names to a specifijic language. These two factors are well-known. Moreover, they are rather obvious. The purpose of this article is to demonstrate that, along with these two factors, which are “external” against the language system, there are also intralingual factors influencing the selection of a letter name. These names tend towards partial unifijication in many languages, that is the tendency to use the same vowel. Such unifijication follows rather distinct and often predictable patterns. The discussion of intralingual factors is hampered by the lack of reliable materials regarding, 1. tradition, and 2. origin of naming short letters. Even in my native Russian, neither I nor any of my learned colleagues are able to satisfactorily answer the questions as to where established Russian short letter names came from and when. Certain essential features allow us to conclude that we are correcting negligent pupils according to the norms of the Latin tradition. Are there other factors that will shed light on the questions and how can we proceed? I became aware of this problem for the fijirst time while studying the material of young Polynesian traditions. Let us consider the Tahitian alphabet as an example. “Creation of the Tahitian alphabet dates back to the early 1800s when English missionaries developed an alphabet on the

about “short” names of letters

71

basis of Latin script containing of 13 letters and named piapa ‘abecedary’ ” (Arakin 1981) (see table below). As is well known, Polynesian languages have remarkably few consonants. Only nine exist in the Tahitian language, including a glottal stop consonant not marked with any symbol. This makes certain oddities in naming of letters even more obvious. Why [pi], [ti], [vi], but [mo], [ro], and [nu] right there, as well as [he], and, fijinally, [fa]? What made such diversity of vowels appear in the names of letters (all the fijive vowels available are “involved” in designation of consonants as its segments)? What is the logic of vowel selection in each case, if any? We cannot explain this oddity by the Polynesian tradition as no tradition exists. Let us assume that in case of [pi], [ti], and [vi] we are facing a direct borrowing from English, but the English pattern is not applied for the designation of the fijive remaining consonants. It can be noted that the pattern is not applied in cases where the prototype names of English letters deviate from the Ci structure, most common in the English language, and have a diffferent structure—[ef], [em], [en], [ar], [eich], that is, the (V)VC structure (a nominal phonetic transcription is quite sufffijicient for the purposes of this article). A certain unifijication of consonant designation took place in the Tahitian language, so that each designation, as opposed to English, has a CV structure. But what kind of unifijication are we talking about if three letters with names containing a similar vowel—[ef], [em], and [en]—turn into [fa], [mo], and [nu]? The Tahitian alphabet Letters Aa Ee Ff Hh Ii Mm Nn Oo Pp Rr Tt Uu Vv

Names of Letters

Pronunciation

A E Fa He I Mo Nu O Pi Ro Ti U Vi

[a] [e] [f] [h] [i] [m] [n] [o] [p] [r] [t] [u] [v]

72

konstantin pozdniakov

Let us consider the aforementioned issues in “prototype” systems, namely the Greek and the Latin systems. To the best of my knowledge, no short names of letters in the Greek tradition are known to have ever existed. Let us have a look at long names of Greek consonants and group them by vowels they contain: – zeta, theta, beta, delta; – gamma, kappa, tau, lambda; – mu, nu; – pi, xi, sigma, psi, phi, khi; – rho. At fijirst glance, the very formulation of the question of vowel quality in these names (for instance, the [a] vowel is present in the names of both letters designating velar stop consonants—gamma and kappa) is unjustifijied: most names of letters designating consonants are borrowed from Phoenician; consequently, it makes no sense to look for logic of naming in the Greek language. However, let us pay attention to the following two points: 1) all the letters added by the Greek (phi, khi, psi) have a unifijied vowel, besides, it is an [i]-vowel; 2) where borrowed from Phoenician, many names of letters got diffferent vowels, and such change seems unmotivated for some of them at fijirst thought. Why *[pe] > [pi], but *[mem] > [mu]? Without knowing the basics of Greek philology, hypothesizing about a linguistic tradition that has been developing for over 2000 years would be absurd. Still, proceeding from a certain experience of research in the fijield of diachronic analogical changes in various languages of the world, I may be so bold as to postulate that the *[mem] > [mu] change could well have taken place by analogy with *[nun] > [nu]. In such a case, names of letters designating nasal sonants would bear a common vowel marker that no other Greek letter name has. If so, vocalic unifijication of certain names of letters, at least letters designating nasal sonants, is present not only in alphabets deriving from Greek but in the Greek alphabet itself as well. Let us note one more detail, which, as we shall demonstrate later in this article, is typical for quite a number of derivative alphabets. The [a] vowel in gamma is not justifijied by a Phoenician origin. Doesn’t the [a] vowel represent a marker of names of letters designating velar consonants already in Greek? For this is the very peculiarity we observe in quite a number of derivative alphabets.

about “short” names of letters

73

With little known about Etruscan letter naming, let us proceed directly to the Roman alphabet. It was precisely the short names of Latin and not Greek letters that served as a source of letter naming in most European alphabets including Cyrillic ones. I was unable to determine whether short Latin names derived from a Greek (Etruscan) source, or developed independently. It is important to know for our further narration that three general types of letter names have developed in the Latin tradition: 1. names with a CV structure, with [e]-vowel, including [be], [ce], [de], [ge], [pe], [te], [ve]; 2. names with a CV structure, with [a]-vowel, including [ha], [ka]; 3. names with a VC structure, including [ef], [el], [em], [en], [er], [es] (Gordon 1973: 30). The life journey of the names of these very letters will be of interest for us. First of all, it is worth noting that any phonetician can easily identify the features determining the phonetic nature of the names belonging to groups 2 and 3: – group 2 includes velar consonants, and, if we set ourselves a task to defijine peculiar features of each group, we should say that group 2 includes unvoiced consonants (group 1 contains the voiced velar [ge]); – there are no stops in group 3; it contains only fricatives and sonants. Let us keep in mind that all the fricatives and sonants are included in group 3, except [h] in group 2. Regardless of the somewhat obscure history of origination of short letter names, rather distinct trends are evidenced in Latin, such as: names of all stops except velars (unvoiced?) ones have a Ce structure; names of velar (unvoiced?) consonants have a Ca structure (that is, compact consonants tend to combine with the most compact vowel!); names of non-obstruent consonants, namely fricatives and sonants, have an eC structure. That was the starting point of all European alphabets, and, as we shall see below, irrespective of the “authorship” and the specifijicity of a language, these purely phonetic features of Latin letter names have been treasured in most new traditions and preserved to the present day.

74

konstantin pozdniakov

Let us have a look at the Spanish system (Ortografía 1999). The letters we are interested in have the following names assigned (since 1803):2

be

ce

de

ge

pe

te

che

efe

ele

eme

ene

erre, ere

ese

elle

hache cu ceta, ceda, zeta, zeda

ka jota

eñe

uve, uve equis ve doble

I griega

Initially, two innovations should be understood. First, the names of the letters were subject to further unifijication. Instead of the Latin opposition of the Ce and the eC structures, less polar Ce and eCe structures are opposed in Spanish. Second, the phonetic “algorithm” as observed in Latin has not changed. It is especially evident in the development of names of new letters. The ll digraph fijirstly representing the [lj]—“fonema lateral palatal de llave, se aricula con la misma pronunciación que la letra y, es decir, como el fonema palatal sonoro de yunque” (Ortografía 1999: 13) is not designated as *[lje] or *[je] or *[aj] in the Spanish tradition but as [elje], that is, similarly to the way fricatives and sonants were designated in the Latin tradition. The same rule applies to designation of the nasal palatal sonant [eñe]. It becomes clear in the light of this rule, why the voiced labial consonant is named [uve / ube] and no more [ve / be] in Spanish—spirantization of the voiced stop took place in Spanish (*b > v), and, consequently, the relevant letter was supposed to acquire the VC(V) structure and not the CV one. Another question arises here: why is the Spanish letter called [uve] and not *[eve]? Perhaps it is the phonetic principle of the generation of letter names that reveals itself even more evidently here: a labial consonant is preceded by a labialized vowel in a letter name. It should be noted that, despite a more consistent phonetic-based unifijication of letter names, the Spanish [ka] is the only letter having a Ca-structure. The case is that the Spanish language had to make a serious

2

 This variant has been used by the Academy since 1803.

about “short” names of letters

75

decision: whether to appropriate the name for [h] inherited from Latin to group 2 (velar) and to preserve its [a]-vowel, or to group 3 (non-obstruent) implying the VC(V) structure. The Spanish judgment worthy of Solomon is an [ache] name instead of *[che] or *[eche]. The [a]-vowel along with the VC structure emerges in Portuguese as well, where the Latin h is matched by [aγa]. However, the Brazilian system preserves the [ka]-name for the velar voiceless stop, while in Portugal the corresponding name is unifijied according to the “standard” principle, having transformed into [ke], as in Russian ke-le-me-ne. A similar principle is observed in the Italian alphabet where voiceless velars are designated as [akka] (the Latin [h]) which is related to the Greek [kappa] name of the voiceless velar stop. The same solution has entrenched in the French alphabet. The name of a letter designating the [h] sound must have a VC structure. As this is not a stop (see French names of other non-obstruent consonants—[ef], [em], [en], [el], [er], [es]), and at the same time, being a name of a velar consonant, it must follow [a]-vowel. Both conditions are fairly met by the name [aš] which is preserved in the French tradition. While considering the French system, further evidence of signifijicance of the phonetic factor for the general letter naming logic, a rather powerful one, in my opinion, would be appropriate to mention here. There are masculine and feminine letters in French. A Frenchman, at least one who lived in the 1800s, would have said: un B, un D, un C, but une M, une F, une S. It seems, according to a mini-poll among native speakers of French, that [l] allows both variants: un L and une L. In Grévisse (1993: 730) we read: “. . . lorsque le nom des consonnes commence par une voyelle, f, h, l, m, n, r, s, il est féminin selon Littré, selon le Dict. gén. et selon l’Acad. (qui donne pourtant les deux genres à f: Un grand F. Une petite f). Cette usage existe encore, mais le masculin prévaut très nettement, notammement parmi les grammairiens et les linguistes d’aujourd’hui”. Why is it so? Perhaps the selection of a relevant article is not focused upon the nature of the sound designated by a letter but upon the very name of the letter: a letter name starting with a consonant (for instance, [de]) is preceded by a vowel (which is nasal in our case) (un), while a name starting with a vowel (for instance, [ef]) is preceded by the article ending with a consonant (une). But what is to be done with le D and la F then? Is it an analogical unifijication on the basis of the indefijinite article? But why do Frenchmen say un A instead of une A? Consistently, a native speaker of French opposes names of letters designating stops, on the one hand, and names of letters designating non-obstruent sounds, on the other hand, by

76

konstantin pozdniakov

gender. Thus, structural opposition of letter names [pe] / [ef] and [ka] / [aš] is anchored in the French language. Is there a phonetic principle in names of letters? One of the most reliable ‘tests’ is checking the names of letters that designate new sounds, that is, sounds that did not exist in the donor language. We have already seen that, for example, the Spanish tradition assigns such names to designate phonemes not present in Latin and that are “aligned” with the phonetic nature of these sounds. Thus, for instance, the letter designating the nasal palatal sonant in Spanish is named [eñə] but not *[ñə]. In Polish, such names emerge as [en’], [es’], along with the inherited [en] and [es]. In Hungarian, we see esz for [s] along with es for [š], eny along with en, and el ipsilon or ely for [j], along with el. An interesting modifijication of Latin letter names is observed in the Russian language. This very point will cause raised eyebrows among most Russianists, for they know, without doubt, that the Cyrillic alphabet came to us from the Greek tradition. And one does not have to know the history of development of the Cyrillic alphabet; a mere comparison of the graphic forms of letters and the sounds designated by the letters will be a sufffijicient proof of this. But where have the Russian short letter names come from? If I take the liberty to repeat, it seems like we do not know whether short names of letters ever existed in Greek. However, we can see almost a complete match of short letter names in Latin and in Russian where such letters designate relevant sounds. Let us satisfy ourselves by comparing Latin and Russian names according to the three abovementioned groups:

Latin Russian

be be

ce ce

de de

ge ge

pe pe

te te

Latin Russian

ef ef

el el

em em

en en

er er

es es

Latin Russian

ha ha

ka ka

ša

šča

ve ve

že

ze

če

Such coincidences cannot be random. Direct borrowing of names from Latin or from some language oriented at the Latin tradition is obvious. But it is not possible to be sure when and how this borrowing occurred.

about “short” names of letters

77

Names of French letters have developed such subtleties that are not to be found in the Russian language: the Russian tradition stands closer to the Latin and not the French one (compare Russian [ge], [e], [že], [ha] with French [že], [œ], [ži], [aš], respectively). As for the Greek tradition, there appear to be no grounds to consider it at all. So, the peculiar features of the Russian names are related fijirst of all to fijive letters (sounds): three of them ([ze], [če], and [že]) are universalized according to the “standard” algorithm, while two others ([ša], [šča]), according to the algorithm applied to velar consonants. Again we see the proof that the peculiarities are related to the sounds that did not exist in the source language or had no separate letters to designate them. There is no verifijiable explanation for the origin of short letter names, names we are using quite often in Russian, and no information is available in encyclopedias, reference books, or consolidated works on the Russian language. The fact is our teachers give the highest grades to those children who make an [e]-vowel after names of most stops and an [a]-vowel after names of velar and sibilant consonants (another display of the “compactness” principle?), and who use the [e]-vowel before names of all sonants and before most of fricative consonants without pronouncing it after the consonant has been defijined. Thus, the contemporary Russian tradition supports the phonetic principles of letter naming that were developed in the Latin language. Some systems eventually tracing back to the Latin system look unusual compared to the rest, however the principle of structural opposition of letter names remains the same. Let us have a look at the Finnish system (Ahonen 2005: 16–17) as an example: Aa Bb Cc Dd Ee

[aa] [bee] [see] [dee] [ee]

Ff Gg Hh Ii Jj

[äf] [gee] [hoo] [ii] [jii]

Kk Ll Mm Nn Oo

[koo] [äl] [äm] [än] [oo]

Pp Qq Rr Ss Tt

[pee] [kuu] [är] [äs] [tee]

Uu Vv Xx Yy

[uu] [vee] [äks] [yy]

Similar to the systems discussed above, one group in Finnish, according to letter names, includes unvoiced velar consonants [hoo] and [koo], and the second group includes sonants and fricatives [äf], [äl], [äm], [än], [är], [s], with [ks] aligned to them. Names of stops (except [koo] and [kuu]) have a Ce structure which developed from earlier CV or Ce. By the way, it is entirely possible that we are dealing with a more orderly unifijication of

78

konstantin pozdniakov

names for velar voiceless consonants, with all the three names containing a deep labialized vowel: [hoo], [koo], and [kuu]. Let us go back to Oceanian alphabets mentioned in the beginning of the article. The Tahitian language contains so few consonants and names of these consonants are so unusual that the easiest explanation naturally suggests itself, namely that the English missionaries based their development of letter names upon their own criteria, unknown to us, which offfer no typological interest, and proceed from the factor designated as “convention” earlier in the article. Eeyore spoke uncompromisingly about this factor: Eeyore had three sticks on the ground, and was looking at them. Two of the sticks were touching at one end, but not at the other, and the third stick was laid across them. Piglet thought that perhaps it was a Trap of some kind “Do you know what this is?” “No,” said Piglet. “It’s an A,” said Eeyore severely. “Christopher Robin said it was an A, and an A it is—until somebody treads on it,” Eeyore added sternly. (Milne 1926)

The missionaries said that it was a [mo] (all letter names are given with long vowels in the Tahitian academic dictionary, particularly [moo]), so a [mo] it was. But the truth is that similar letter name peculiarities are evidenced in languages of the Austronesian family not belonging to the Polynesian group and developing their own alphabets based upon the Latin tradition independently (as it may be suggested) from the convention that was generated on the Tahiti island. Let us consider letter names that have settled in Austronesian languages of another group, namely Drehu and Nengone, which are spoken on the islands of the Commonwealth. Actually, it was exactly the strange letter names in the Drehu language recorded in 1996 that, according to Wamo Haocas, a native speaker of this language, gave an impulse to raising the issue of the phonetic nature of letter names in alphabets.3 Regarding the Drehu language, my records, for the most part, match comments given by specialists, however, minor variances exist that may be caused by inconsistencies in the records or simply an unestablished

3

 For the purpose of this article, these recordings were checked by one of the most competent authorities in the fijield of Austronesian languages, Claire Moyse-Faurie, who redirected my questions to specialists in the Drehu and Nengone languages (Jacque Vernodon, Emma Troopy, Tila Vaitiko, Julie Xmae). My acknowledgements to them all.

about “short” names of letters

79

Names of Letters in Drehu Labials Dentals I I I I E E A A A O O U

Stops

Voiceless Voiced

Fricatives Sonants Stops Fricatives Fricatives Voiceless Voiced Sonants Fricatives Sonants Sonants

pi bi vi

Velars

Labialized Velars

ti di (dri [ji] ) ci ([ši]) (nyi [ñi]) ze sa

fa wa

Palatals

ða la ro

ke he xa ga ([γa]?)

wa (qa [xwa])

ja ([y] in ny)

mo nu

Comments to the table: – Consonants are grouped by orders in columns and by series in rows. This two-dimensional table does not allow for the inclusion of a third feature—a vowel quality. For this reason, the vowel feature and the series feature are combined in relevant rows where possible. – The table does not include the glottal stop consonant. Names of the glottal stop in Oceanic traditions give an example of a purely conventional factor afffecting letter-naming. Thus, in Wallis the glottal stop is called fakamoga (faka ‘to do; causative prefijix’; moga ‘Adam’s apple’, while its name in the Futu language is apositolofiji (derived from the French apostrophe) or gā fakatu’u, for the reason that pronouncing of a vowel is preceded by a breath catch on the glottis level (Claire Moyse-Faurie, personal communication).

norm. They may also be the result of dialect diffferences related to implementation of certain Drehu phonemes. Below, in brackets I provide the data obtained for the cases in which diffferences or adjustments in my records appear important. It is easy to see that the “exotic” system of the Tahitian alphabet is fully represented in the alphabet of Drehu, one of the languages spoken on New Caledonia! At the same time, not including the glottal stop, Drehu contains 22 consonants, unlike Tahitian with 8 only. It can be concluded that it would be easier to defijine system attributes using the material of Drehu. The question at issue is whether these attributes can be defijined at all. There are a few rules and some conclusions as to possible trends worth mentioning: 1. All names of consonants have a CV and not a VC structure. That is, in the languages considered, structural unifijication of letter names according to the “main” variant took place. 2. All the letters having a “standard” Ci-structure in English, preserve their names.

80

konstantin pozdniakov

3. All the names with [a] identify non-obstruent consonants, designating fricatives and sonants. 4. The same applies to names with [o]- and [u]. To put it another way, non-front vowels are an attribute of names of non-obstruent consonants. 5. All stops except the velar [k] bear names with [i]. 6. Except stops, [i] inherited from the standard English denomination of letters is present only in the names of palatal consonants, particularly [ši] and [ñi]. As likely as not, it is the palatal consonant that determinates the [i] for non-obstruent sounds. 7. With the exception of [ze], which might have derived directly from the English [zed], [e]—is the particular marker of both names for velars, [ke] and [he]. 8. Perhaps the names [ro] and [nu] are direct quotations of the Greek tradition, for the Latin tradition did not fijit in here due to structural unifijication of all letter names (CV structure). 9. The name [mo] instead of [em] (Latin) or [mu] (Greek) might have been determined by the labial nature of the consonant, exactly as the palatality of [ši] and [ñi] defijines inclusion of the vowel [i] into the name of the letter. To conclude, the description of the Drehu alphabet requires the application of many rules, despite the condition that the CV structure demonstrates an obvious trend to further unifijication of Drehu letter names as compared to the English alphabet. As for the alphabet of Nengone, yet another language of New Caledonia to which the principles of letter-naming defijined for Drehu and Tahitian are also generally applied, (including the CV structure, such names as [mo], [nu], [ro], [ ja], [wa], and a number of other signifijicant attributes), so many more special rules are necessary that this most strange system defijies satisfactory description. Below are letter names of this language grouped by vowels: – Vowel I: [pi], [ti], [bi], [di], [vi]. – Vowel E: [ke], [he], [ge], [ce], [se]. – Vowel A: [fa], [xa], [wa], [la], [ja], [za]. – Vowel O: [ro], [mo], [yo]. – Vowel U: [nu]. Rules 5 to 7 established for Drehu do not match the description of the Nengone alphabet. Instead, other rules are required, such as:

about “short” names of letters

81

– [e] is typical for all velar and all voiceless palatal consonants, that is, for most of the compact consonants (see the Russian grouping of voiceless velar and palatal consonants—with a diffferent vowel quality—in [ka], [ha], [ša], and [šča]). – It is not known why the palatal sonant has [o] in its name ([jo]). When the number of rules approaches the number of consonant names these rules describe, one should rather speak about a random character of letter names. But somehow, it was these and not some other letter names that became accepted in Nengone! It is established that these alphabets were developed by English missionaries in the nineteenth century (Lenormand 1953). How were the changes efffected and how did it come to be that Nengone letters names developed in this manner and not some other system? Transformation of English names in Tahitian, Drehu and Nengone seem radical and inconsistent in appearance. Even these “exotic” systems demonstrate two factors defijinitely deserving attention. Why does unifijication follow the yo-ke-le-me-ne pattern and not, let us say, yo-ka-la-ma-na? Before formulating my assumption, let us return to the French language. French schoolchildren are taught the following: there are “names of letters”, and there is “pronunciation (!) of letters”. For example, the name of the letter is “em”, while the pronunciation of the same letter is [m?]. (By the way, this concept is quite intriguing: should we implement it, we would be able to “legalize” the diffference between [em] and [me] by defijining the former as the name and the latter as the pronunciation of the letter. But what would represent the name of the sound [m] better, in such a case?!) Moreover, there is a third variant in the French language, which has no “label”. The French word femme ‘woman’ consists of one syllable and ends with the consonant [m]. But the poetic tradition requires that this word contains two syllables. Which vowel is then pronounced when this word is included in a poetic text? Neither [e] nor [ε], but [œ] is pronounced. And if a Frenchman has to articulate this word clearly (for instance, when someone cannot understand this word on the phone), the Frenchman will say, [fa-mœ]. What are we dealing with here, when it is neither the name nor even the pronunciation of the letter? The French slang used by young people allows for such a feature as metathesis, just like many other types of slang. Here, femme turns into mœf, and fête ‘holiday’—into tœf. That is, the “mute” fijinal vowel is implemented as [œ]. Besides, the metathesis of the word flic ‘policeman’ where no “graphic” vowel is present either, comes out as kœf, not kif or kef!

82

konstantin pozdniakov

All this is absolutely impossible in the Russian language. Why? A conservative assumption would be that in most languages—or, to put it more precisely, in many languages (considering Drehu and Nengone, one should not declare universals)—there is a tendency to use either shwa or, if there is no shwa in the system, a vowel standing closest to shwa as part of the “pronunciation” of letters. In the Russian language, the closest vowel to [ə] is [e]. This is where yo-ke-le-me-ne comes from. The French language, distinct from Russian, has an order of front labialized vowels and the closest vowel to [ə] is [œ]. This is where kœf came from. I believe that even the fragmentary material provided here indicates that the fijield of letter names can be studied as a system of changes by diachronic analogy and would benefijit from the continued attention of linguists.4 References Ahonen, L. 2005. Sound good—kuulostaa hyvältä. Helsinki: Finnish Literature Society. Arakin, V.D. 1981. The Tahitian language. Moscow: Nauka. Gordon, A.E. 1973. The Letter Names of the Latin Alphabet. Classical Studies, Vol. 9. Berkeley: University of California Press. Grevisse, M. 1993. Le bon usage. Thirteenth Edition. Paris—Louvain-la-Neuve: Duculot. Lenormand, M.-H. 1953. Notice sur la graphie du Lifou et l’histoire des alphabets loyaltiens. Études mélanésiennes 7: 15–18. Milne, A.A. 1926. Winnie-the-Pooh. Dutton Juvenile. Ortografía de la Lengua Española 1999. Edición revisada por las Academias de la Lengua Española. Madrid: Real Academia española. Plato 1875. Cratylus. In The Dialogues of Plato, Volume II. Translated by Benjamin Jowett. Oxford: Oxford Clarendon Press.

4

 I would like to thank the anonymous reviewers for their helpful comments.

EARLY ADAPTATIONS OF THE KOREAN SCRIPT TO RENDER FOREIGN LANGUAGES Sven Osterkamp Usage of the Korean alphabet, han’gŭl 한글, is for the most part restricted to write Korean itself and so far there has not been any other language it is commonly associated with as its default script. How deeply rooted this one-to-one correspondence is can neatly be illustrated by the appellation hanguru-go or even simply hanguru (ハングル[語], i.e. ‘han’gŭl [language]’) currently in use in Japan besides various others to refer to the modern Korean language. Little time had passed however since its creation in the 1440s before han’gŭl was used to render a variety of foreign languages, often in order to serve didactic purposes. Especially at the so-called Sayŏgwŏn 司譯院, or Bureau of Interpreters, it was widely employed to gloss the pronunciation of neighboring languages such as the following. The earliest date each language can be demonstrated to have been rendered in han’gŭl is provided in brackets, dates of non-extant texts are starred: – Chinese (*1450s: Sasŏng t’onggo 四聲通攷) The Sasŏng t’onggo is one in a series of early rime dictionaries to which also belong the contemporary Hongmu chŏngun yŏkhun 洪武正韻 譯訓 (1455), a Korean adaptation of the Míng period rime dictionary Hóngwǔ zhèngyùn 洪武正韻 (1375), and the Sasŏng t’onghae 四聲 通解 (1517). Besides we fijind for instance a Chinese–Korean dictionary Yŏgŏ yuhae 譯語類解 (1690, addendum 1775) and fully glossed textbooks like Pŏnyŏk Nogŏltae 飜譯老乞大 and Pŏnyŏk Pak t’ongsa 飜譯朴通事, both of which were fijirst published no later than 1517 and enjoyed several revisions in the 17th and 18th centuries. – Japanese (1492: Irop’a 伊路波) Most important here apart from the brief primer Irop’a is the Ch’ŏphae sinŏ 捷解新語 series of bilingual textbooks with three major installments featuring transcriptions in han’gŭl (1676, 1748 and 1781). The only dictionary published in print is the late 18th century Waeŏ yuhae 倭語 類解.

84

sven osterkamp

– Ryūkyūan (1501: “Ŏŭm pŏnyŏk” 語音翻譯) Among the languages taught at the Bureau Ryūkyūan was clearly secondary. In 1437, for instance, students of Japanese were made to take up the study of Ryūkyūan as well, but sources are silent on the teaching materials used. While not compiled at the Bureau, han’gŭl transcriptions of interest are also found in the early 19th century P’yohaerok 漂海錄 besides the above-mentioned “Ŏŭm pŏnyŏk”, an appendix to Haedong chegukki 海東諸國紀 (1471). – Manchu (*1704: Ch’ŏngŏ nogŏltae 清語老乞大) The extant works on Manchu mostly date from the second half of the 18th century. Among the textbooks we fijind revisions of Ch’ŏngŏ nogŏltae (1765), Samyŏk ch’onghae 三譯總解 (1774), P’alsea 八歳兒 and Soaron 小兒論 (both 1777; all four previously published in *1704); to these we can add two dictionaries, namely Tongmun yuhae 同文類解 (1748) and Han Ch’ŏngmun’gam 漢清文鑑 (1779?). – Mongolian (*1737: Ch’ŏphae mongŏ 捷解蒙語) All extant works on Mongolian date from the year 1790: Revised editions of the textbooks Ch’ŏphae mongŏ and Mongŏ nogŏltae 蒙語老乞大 (previously published in *1741, *1766) as well as a new edition of the dictionary Mongŏ yuhae 蒙語類解 (fijirst edition *1768). To a greater or lesser extent all of these languages difffer from Korean in respect to their inventory of sounds as well as their phonotactical rules to combine these sounds. Apart from a certain conservativeness in writing—namely in that letters that dropped out along with the corresponding phonemes in Korean over the course of its history were retained (e.g. ㅿ)1—this lead to a range of innovations, or to put it diffferently, to a number of deviations from the prototypical usage of the script as used to write Korean.

1

 While McCune-Reischauer romanization is used here for Korean names and titles, it is insufffijicient for the transcription of the entire repertoire of han’gŭl letters, i.e. including those discussed in the following. Therefore, an expanded and somewhat modifijied version of Yale romanization has been adopted instead for these purposes, in which the basic consonants are transcribed the same, but which also provide transcriptions for all the special letters used for Chinese and other languages (on which see tables 1, 2 and 5). Concerning vowels (cf. table 8), and likewise is written instead of Yale and respectively, which is arguably a more intuitive choice of symbols, but also helps avoiding ambiguities involving ㅱ. Finally, is used instead of , but is written for the modern monophthongs deriving historically from the diphthongs written here; Yale has for these instead, regardless of their actual phonetic realization at the relevant time.

early adaptations of the korean script

85

In the following some of the more important deviations from common usage will be outlined, largely drawing upon works for the instruction in Chinese, Japanese and Manchu at the Bureau of Interpreters, with some additional materials of later times unrelated to the Bureau. Additional Letters The dividing line between (compositionally complex) letters and cases involving diacritical marks is hard to draw at times. For practical purposes we will consider a circle placed outside the syllable block as a diacritical mark, whereas one placed below or next to a consonant letter and integrated into the syllable block proper as part of a complex letter. Type A: Doubled base letter Since the days of its creation han’gŭl included several letters unnecessary to write Korean proper, but indispensable when it came to reducing early Sino-Korean and especially Chinese to writing. As early as 1448 a rime dictionary for a prescriptive system of Sino-Korean character readings was published in form of the Tongguk chŏngun 東國正韻. The system was relatively short-lived and replaced before the end of the century, in all likeliness mainly due to fact that it contained distinctions not to be found in Korean (and thus likewise not in Sino-Korean as it was in actual use either). It is exactly these distinctions which gave rise to the fijirst type of additional letters: those formed by means of doubling a given base letter. In order to imitate the tripartite distinction between voiceless unaspirated (quánqīng 全清), voiceless aspirated (cìqīng 次清) and voiced unaspirated obstruent initials (quánzhuó 全濁) found in Chinese, distinct letters for voiced obstruents had to be added. This was done by doubling the letters used for initially voiceless unaspirated obstruents as shown in table 1 together with the regular letters for aspirates (only the pronunciation in word-initial position if used for Korean proper is indicated for the base letters).2

2  It is commonly believed that the letter already stood for [h] at the time of the alphabet’s creation—that is, just as it does today—, but for various reasons an original [x] seems likely, which underwent lenition only later on.

86

sven osterkamp Table 1. Base and derived letters

base letters (voiceless)

derived letters (voiced / glottalized)

derived letters (aspirated)

ㄱ

ㄷ

ㅂ

ㅈ

ㅅ

ㅎ

k [k]

t [t]

p [p]

c [ʦ]

s [s]

h [h] ([x])

ㄲ

ㄸ

ㅃ

ㅉ

ㅆ

ㆅ

g [g]

d [d]

b [b]

j [ʣ]

z [z]

x [ɣ]

kk [k’]

tt [t’]

pp [p’]

cc [ʦ’]

ss [s’]

hh [h’] ([x’])

ㅋ

ㅌ

ㅍ

ㅊ

kh [kʰ]

th [tʰ]

ph [pʰ]

ch [ʦʰ]

Another set of such doubled letters to indicate voiced consonants in Chinese will be dealt with below under type C. Outside Sino-Korean or renderings of Chinese the derived letters given above are commonly considered to render geminate or glottalized obstruents just as in current usage. As overt borrowings from Chinese were accompanied by their respective characters in early printed text, their presence would then have triggered the derived letters’ voiced reading as opposed to an interpretation as geminates. In passing a further letter created in the same way deserves mentioning, namely ㅥ, which is found both in Korean texts and in foreign language materials. As ㄴ already renders a voiced /n/, it is obvious at fijirst sight that here the doubling of the base letter can only be understood as standing for gemination or long articulation. The attestations of provided by Nam (1997: 354)—in all of which it renders /nn/ as the result of assimilation of stem-fijinal /h/ to the initial /n/ of an attached sufffijix— confijirm this. At least in the materials on Japanese there are then also cases of clearly rendering a long nasal. In the 1676 print of the Ch’ŏphae sinŏ for instance we fijind Japanese nan=no ‘what? + attributive’ written as 나 (X/32b). The exact motivation here and elsewhere is somewhat unclear, it is safe to say however that in the works on Japanese it is always the geminate reading of the doubled letters that is intended, never the voiced one as with Chinese.

early adaptations of the korean script

87

Table 2. Letters with circle beneath base letters (bilabial)

derived letters (labiodental)

ㅂ

ㅍ

ㅃ

ㅁ

p [p]

ph [pʰ]

b [b]

m [m]

ㅸ

ㆄ

ㅹ

ㅱ

f [f]

fh [f(ʰ)]

v [v]

w [ʋ] ([w])

Type B: Base letter plus circle Unlike bilabials, labiodentals are lacking altogether in Korean,3 while the variety of Chinese recorded in the above-mentioned materials had a stable phonemic opposition of bilabials versus labiodentals (zhòng chúnyīn 重唇音 ‘heavy lip sounds’ versus qīng chúnyīn 輕唇音 ‘light lip sounds’ in traditional terminology). A truthful transcription thus required means to diffferentiate the two types of labials. In view of the fact that the Chinese labiodentals developed out of earlier bilabials it is unsurprising to fijind the letters for labiodentals being similarly derived in a regular fashion from the already existing set of letters for bilabials: simply by adding a circle beneath the base letter (see table 2). The distinction between ㅸ and ㆄ appears to be a theoretical one and is, in all likeliness, merely etymologically motivated: /f/ from /p/ could be written as ㅸ, while /f/ from /ph/ would then give ㆄ. In fact however, was not even used in the earliest works like Hongmu chŏngun yŏkhun or Sasŏng t’onggo. In the explanatory notes to the latter (item 4), which are quoted and thus preserved in the later Sasŏng t’onghae, its author Sin Sukchu 申叔舟 (1417–1475) explicitly mentions the non-existence of such a distinction in contemporary Chinese and his decision to use throughout. Some examples of how the base letters as well as the derived ones were used to render Chinese are provided in table 3 (taken from Hongmu chŏngun yŏkhun XI, rime 5; the initials in the modern standard pronunciation given here in Pīnyīn for the sake of comparison are straightforward, but note: b [p], p [pʰ]).

3

 Maybe apart from a short-lived phoneme—the pronunciation of which in intervocalic position is usually considered to have been bilabial [β] but which might perhaps have been labiodental [v]—as the result of lenition of original /p/, intervocalically [b].

88

sven osterkamp Table 3. Examples for the two labial series

character

bù 布

pù 鋪

bù 步

mù 暮

bilabial initials

pu 부

phu 푸

bu 뿌

mu 무

character

fù 付

fù 父

wù 務

labiodental initials

fu 

vu 

wu 

( fhu )

In the Irop’a of 1492 both and even are used to render the bilabial fricative of Japanese, but in a rather unsystematic fashion and in fact only in one portion of the work. In another portion and also are found instead. Syllabic /u/ is further rendered here as  once, which is reminiscent of several early European transcriptions of the Japanese syllabary: To name but a few cases, Meister (1692: plate “Das Japansche A.B.C.”) writes “wu”, Müller (1694: “Syllabarium Japanicum geminum”) similarly “wou” and Overmeer Fisscher (1833: plate “Het Japansch Alphabet met zogenaamde Katakane letters”) has “woe”. However the main bulk of works on Japanese used at the Bureau, all published during the course of the 17th and 18th centuries, makes no use of the set of letters . The only early source on Ryūkyūan on the other hand, the “Ŏŭm pŏnyŏk” of 1501, uses for word-initial w- on several occasions (all for the fijirst person pronoun wan), but none of the other letters.4 Out of the four derived letters only ㅸ occurs regularly in works on languages other than Chinese. Labiodental /f/ in Manchu is exclusively rendered as in works using what might be termed a narrow transcription (i.e. making use of special letters and diacritical marks, on the latter of which see below), while those using a broad one resort to the best match available in Korean proper, namely ㅍ. The latter is also the default way to incorporate foreign labial fricatives in borrowings to the present day. In times postdating the Bureau’s activities we fijind further cases of employing circles to indicate articulations that are diffferent from, yet still similar to some extent to that of the base letters they are attached to. Unlike the early cases with superscript circles, the transcriptions of English found in the quatrolingual vocabulary Ahakp’yŏn 兒學編 of 1908

4

 The variety of Ryūkyūan reflected in this source is usually considered to have retained /p/ as a stop, either entirely so that labial fricatives did not yet exist at all (thus e.g. Hattori 1979: 89), or at least partially, so that the newer fricative articulation co-existed with the older plosive one (e.g. Tawata 1979: 27).

early adaptations of the korean script

89

Table 4. Transcriptions of English words in Ahakp’yŏn derived letter

sound value

English

transcription

page

ᅋ

[f]

knife

na-i-fij 나이

27a

ᅄ

[v]

have

hɛ-vɨ 해

51a

thin

þin 

50b

mother

mo-þə 모

1a

throat

þɨ-lo-thɨ 로트

5a

ᅈ

[θ] ~ [ð]

ᅂ

use a set of letters derived by placing a circle to the left of the base letter. Labiodental /f/ and /v/ are regularly rendered by (or ) ᅋ and () ᅄ respectively, while the interdentals are transcribed as either () ᅈ or less commonly () ᅂ without a clear-cut voicing distinction. The derived letters and some sample transcriptions found in this work are presented in table 4.5 Type C: Mirrored letters Sibilants in Middle Korean are generally deemed to have been alveolar ([s, ʦ] &c.), possibly with alveopalatal allophones ([ɕ, ʨ] &c.) before front vowels or palatal glides. Retroflex articulation seems not to have been known in Korean at any time. Just as today’s standard language the variety of Chinese taught at the Bureau further distinguished a series of retroflex sibilants ([ʂ, tʂ] &c.) however, which stood in opposition to the alveolar—or possibly dental, if one follows Sin Sukchu’s explanatory notes to his Sasŏng t’onggo—and alveopalatal ones. In an attempt to cope with this situation a way of distinguishing between retroflexes and the rest was devised based on the symmetric nature of the sibilant base letters: prolonging the stroke pointing to the lower right indicated retroflex articulation, while doing the same with the one pointing to the lower left indicated absence of such articulation. See table 5 for the letters in question together with their sound values, and

 Both ᅂ and ᅈ are already found in Waeŏ yuhae, needless to say however not to render interdentals, but rather as clusters of nasal plus obstruent, i.e. and . Occuring each only once (I/45a, I/12a), they are undoubtedly mere misprints for and respectively. 5

90

sven osterkamp Table 5. Mirrored sibilant letters

base letters modifijied base letters ([post-]alveolar)

mirrored letters (retroflex)

ㅅ

ㅆ

ㅈ

ㅊ

ㅉ

ᄼ

ᄽ

ᅎ

ᅔ

ᅏ

s [s, ɕ]

z [z, ʑ]

c [ʦ, ʨ]

ch [ʦʰ, ʨʰ]

j [dz, ʥ]

ᄾ

ᄿ

ᅐ

ᅕ

ᅑ

sr [ʂ]

zr [ʐ]

cr [tʂ]

crh [tʂʰ]

jr [dʐ]

table 6 for illustrations as to how these ten letters were used to transcribe minimal pairs in Chinese (taken from Hongmu chŏngun yŏkhun IV, fijirst tone, rime 11; the initials in Pīnyīn are as follows: x [ɕ], j [ʨ], q [ʨʰ]; sh [ʂ], zh [ʈʂ], ch [ʈʂʰ]). As far as the basic letter shapes are concerned, parallels or similarities between the Korean and Tibetan scripts have been noticed since the early 19th century, thus for instance by Abel-Rémusat (1820: 83f.), Siebold (1832fff.: VII, 14), Wall (1840: 254fff.) and others. It is also possible however to fijind parallels between the two scripts on another level, namely in at least one of their approaches of deriving new letters from already existing ones. In Tibetan we thus fijind a set of letters devised to render retroflex consonants occurring in Sanskrit, all of which are formed by mirroring their non-retroflex counterparts (see table 7; the inherent vowel is ignored in giving the letters’ respective sound values). The corresponding ‘Phagspa letters are also provided for reference.6 In view of the influence the ‘Phagspa script as a descendant of the Tibetan script is deemed to have had on han’gŭl (cf. especially Ledyard 1966 [1998]) it is suggestive to see the same principles applied here as well. At least the letters for /ṭ, ṭh, ḍ, ṇ/ used to write the corresponding sounds in Sanskrit words, as attested in the hexaglot Jūyōng guān 居庸關 inscription of 1345, were created by mirroring the base letters for /t, th, d, n/. One can only surmise that ‘Phagspa served as a model for han’gŭl here—the analogy at least is perfect.7 6  ‘Phagspa sh(a) ꡚ was used to transcribe both Sanskrit ś [ɕ] and ṣ [ʂ] (Higuchi 2001: 731), while what appears to be its mirrored version at fijirst glance is actually the unrelated letter ꡖ, variously claimed to represent either a glottal stop [ʔ] or a phonetic zero. 7  Incidentally Ledyard (1966: 341 [1998: 411]) even reproduces the mirrored for /ṇ/ in a table taken from Hope (1957: 151), but the following comment leaves no doubt that at

early adaptations of the korean script

91

Extension of Possible Syllable Block Structures Middle Korean had seven vowels /a, ə, o, u, ʌ, ɨ, i/, most of which could be followed by an additional palatal glide, the fijirst four also preceded by one. Alternatively /a, ə/ could also be preceded by a labial semivowel. While the letters now have a vertical baseline, have a horizontal one. (nowadays obsolete) consists merely of a single dot, in terms of shaping behavior of syllable blocks however it equals the vowels with horizontal baseline. When two vowel signs with vertical baseline occur together, the second is always to render the offf-glide /y/ (phonetically [j]) in a falling diphthong of the type /Vy/. Vowel signs with horizontal baseline on the other hand do not normally co-occur within a single syllable block. In combinations of vowel signs with horizontal and vertical baselines they only occur in exactly that order. All structures described so far are summarized in table 8.8

least at the time of writing he was not aware of the existence of mirrored letters in ‘Phagspa: “As for the PP [= ‘Phagspa; S.O.] letter that Hope labels “(Ṇ),” I may be making a bad blunder, but I have never seen this letter in other PP lists. It is not even in Hope’s list in his other publication (1953). Where did this come from?“ (Ledyard 1966: 343 [1998: 413]). It is certainly true that the letters for the retroflexes are indeed somewhat underrepresented in the literature on the script—arguably as most studies focus on either Mongolian or Chinese texts in ‘Phagspa script, in neither of which the letters in question ever occur— and they were likewise not included in the early lists of ‘Phagspa letters, such as those found in Měnggǔ zìyùn 蒙古字韻 (1308), Fǎshū kǎo 法書考 (1334), Shūshǐ huìyào 書史 會要 (1376) or Yuánshǐ lèibiān 元史類編 (1706; based on the preceding work). Now, to answer Ledyard’s question: Hope’s “Ṇ” ultimately goes back to its occurance in the Jūyōng guān inscription, part of which he quoted in 1953 in “Schedule IV. A hPhags-pa Sanskrit Text” noting explicitly in the beginning “that reversed N is used for Ṇ, as in Tibetan.” One is left to wonder, however, why Hope (and a number of earlier and later scholars with him) did not notice the other mirrored letters as well, as in fact even the partial quote in “Schedule IV” alone also features mirrored for /ṭ/ for instance – not to speak of the longer collated version in all six scripts provided already by Wylie (1870) or the facsimile of the relevant portion found shortly after in Yule (1871, I: plate following page 28). Finally, owing to Bonaparte’s magnifijicent Documents de l’époque mongole (1895) a reproduction of the original inscription had also long been readily available. Clauson (1959: 304), one of Ledyard’s other sources on the ‘Phagspa script, in fact even touches upon transcriptions of Sanskrit and specifijically the Jūyōng guān inscription as well, stating clearly: “The other problem, that of representing the cerebral sounds, was solved, as it had been in Tibetan, by reversing the corresponding dental letters.” For more recent inventories of ‘Phagspa letters, no longer excluding the mirrored ones, see for instance Junast/Yáng (1984: 381) or Higuchi (2001: 729). 8  The explanation of the script in Hunmin chŏngŭm 訓民正音 (haerye-bon 解例本 of 1446; 23a) also provides for syllables such as /kyʌ/  and /kyɨ/  (not to be confused with /kʌy/  and /kɨy/ 긔), which however did not occur in Middle Korean. Note the unusual arrangement here of vowel signs with vertical and horizontal baseline in that order.

92

sven osterkamp Table 6. Examples for the two sibilant series

character

xiān 先

xián 涎

jiān 煎

qiān 千

qián 前

non-retroflex initials

syen 

zyen 

cyen 

chyen 

jyen 

character

shān 羶

chán 鋋

zhān 氈

chǎn 燀

chán 蟬

retroflex initials

sryen 

zryen 

cryen 

crhyen  jryen 

Table 7. Mirrored letters for retroflexes in Tibetan and ’Phags-pa ཏ

ན

ཤ

ꡉ

ꡊ

ꡋ

t(a) [t]

th(a) [tʰ]

d(a) [d]

n(a) [n]

sh(a) [ɕ]

ཊ

ཋ

ཌ

ཎ

ཥ

ꡩ

ꡪ

ꡫ

ꡬ

mirrored letters (retroflex)

ད

ꡈ

base letters (non-retroflex)

ཐ

ṭ(a) [ʈ]

ṭh(a) [ʈʰ]

ḍ(a) [ɖ]

ṇ(a) [ɳ]

ṣ(a) [ʂ]

Table 8. Possible vowel combinations in Middle Korean S(y) / (w)

V

-S

V

Vy

yV

yVy

wV

wVy

a

aㅏ

ay ㅐ

ya ㅑ

yay ㅒ

wa ㅘ

way ㅙ

ə

əㅓ

əy ㅔ

yə ㅕ

yəy ㅖ

wə ㅝ

wəy ㅞ

oㅗ

oy ㅚ

yo ㅛ

yoy ㆉ

uㅜ

uy ㅟ

yu ㅠ

yuy ㆌ

ʌ

ʌㆍ

ʌy ㆎ

ɨ

ɨㅡ

ɨy ㅢ

i

iㅣ

o (y)

u

(y)

early adaptations of the korean script

93

Table 9. Examples for extended syllable block structures character

Manchu transcr.

piāo 飃

piyao

yuè 月

yuwei

dǒu 斗

deo

I/2b

phiyao 

I/16b I/7a

nyuy 

I/6a

yuəy 

I/3b

ᡩᡝᠣ

nioi

liyang 

ᠶᠥᡝᠢ

nǚ 女

page

liu 

tɨu 

I/4b

ᠨᡳᠦᡳ

lio

ᠯᡳᡉ

liǔ 柳

ᡦᡳᠶᠠᠣ

liyang

ᠯᠶᠶᠨᠢᠠ

liáng 涼

Korean transcr.

This framework is now considerably expanded when it comes to transcriptions of Chinese and, to a lesser extent, also Manchu, Mongolian and Japanese. We therefore fijind examples like those found in table 9 taken from the late 18th century Han Ch’ŏngmun’gam with syllable blocks arranging vowels with vertical and horizontal baselines in that order (, ), containing two vowels with vertical baseline of which not the second but the fijirst is (), combining the fijirst two structures () or arranging two vowels with horizontal baseline () from top to bottom. Also there are cases which are not special as far as they arrangement is concerned, but which still feature constellations that do not normally occur in Korean, such as tautosyllabic . As these are not direct transcriptions of Chinese but rather re-renderings of the Manchu transcriptions found in the Qīngwénjiàn 清文鑑 as the model of the Han Ch’ongmun’gam we provide these original transcriptions as well. As far as vocalism and thus the structure of syllable blocks in this respect is concerned the materials on Japanese yield not exactly much apart from the frequent appearance of  to render part of the stem of the

The prescriptive Sino-Korean readings in Tongguk chŏngun furthermore allowed /yuyə/ in closed and /yuyəy/ in open syllables. Later Sino-Korean at least also features the latter in the reading of some characters.

94

sven osterkamp

verb mōsu ‘(humble for:) to say’. Issues become more interesting when it comes to initial consonant clusters. While Middle Korean had its share of possible clusters, such as ㅳ, ㅵ and so on, combinations of nasal plus obstruent for instance were not valid in syllable onsets. It is exactly in this way however that the formerly pre-nasalized obstruents of Japanese were perceived, and likely also actually pronounced. Thus, to render these obstruents in an adequate fashion, initial clusters consisting of a nasal followed by a homorganic obstruent were frequently used in the teaching materials for Japanese, such as ㅦ for /d/ [nd], ㅮ for /b/ [mb] and so on. Diacritical and Punctuation Marks Diacritical marks can be subdivided into two classes: those whose foreign origin is evident at fijirst sight (type B below) and those for which this does not necessarily hold (type A). The latter category comprises only such diacritics which are so simple in shape—mere dots or short strokes, circles at best—that any claim to a foreign origin sufffers from the possibility of independent invention. Type A: Dots or short strokes and circles Circles serve as a kind of all-purpose diacritic in han’gŭl to indicate a reading diffferent from the default norm. In materials on Manchu and Mongolian it is used to indicate syllable-fijinal /l/ (as opposed to /r/; see below), to keep apart Manchu u and ū and in various other ways. (Compare also the above-mentioned derived letters consisting of a base letter plus a circle next to or beneath it. As these are integrated into the syllable blocks instead of being placed outside of them, it seems inappropriate to regard them as diacritical marks at par with those discussed here. Graphically however they use the same means.) The tone marks (pangjŏm 傍點 ‘side dots’) used also for Korean proper in early texts usually took the shape of short strokes, but they were obviously created after Chinese models—the same indeed as gave birth to diacritical marks in Japanese writing to be discussed presently. Somewhat less certain but still quite probable is the assumption of Chinese influence when it comes to punctuation marks. Korean was originally written without spacing as was Chinese and also Japanese, but Mongolian and consequently also Manchu did make use of spacing. Whenever these were written in transcription only the need

early adaptations of the korean script

95

Table 10.  in Han Ch’ŏngmun’gam versus spaces in Qīngwénjiàn

šun, biyai yabure de teisulebufiji dalibuha (be, šun jembi sembi). „(it is called šun jembi when) the sun is covered, the moon in its motion having met it“ (the portion in round brackets is left out in Han Ch’ŏngmun’gam)

ᡩᡴᠠᠯᡳᠪᡠᡥᠠ ᠪᡝ᠈ ᡧᡠᠨ ᠵᡝᠮᠪᡳ ᠰᡝᠮᠪᡳ ᠉ ᡧᡠᠠ᠈ ᠪᡳᡳ ᡳᠪᡠᡵᡝ ᡩᡝ ᡨᡴᡳᠰᡠᠯᡝᠪᡠᡶᡳ

defijinition of rìshí 日食 ‘solar eclipse’:

러 부 ˎ 다 리 부 하

슌ˎ 뱨ˎ 야 부 러ˎ 더ˎ 테 수

arose to retain the information on segmentation provided by spaces in the original scripts. This was the case for instance with the longer defijinitions in the Han Ch’ŏngmun’gam. These were adapted from those found in Manchu in the Qīngwénjiàn, but are given here in han’gŭl only. Wherever the original Manchu has a white space, the Korean transcription adds a comma-like under the lower right corner of the preceding syllable block. Compare Table 10 which reproduces one sample defijinition from the Han Ch’ŏngmun’gam (I/3b) together with its Manchu source. The same kind of short stroke was also used in other works, an interesting example being the multilingual vocabulary Pangŏn yusŏk 方言類釋 (manuscript fijinished in 1778, but never actually printed). Here all foreign languages except for Chinese are given in han’gŭl transcriptions only, just as was done in the various closely related yuhae dictionaries. The short stroke as a punctuation mark is put into use not merely for Manchu and Mongolian as would be expected, but its use is extended to include Japanese as well—which originally does not use any kind of spacing.9 Similar devices to segment Japanese sentences into smaller units, often going below the level of words, were introduced for didactic purposes in Chunggan kaesu Ch’ŏphae sinŏ 重刊改修捷解新語 (1781) as the last 9  As an anonymous reviewer pointed out, the alternation of Chinese characters and kana syllabograms in Japanese writing can to some extent fulfijill a function quite similar to that of spacing. This is however little more than a rule of thumb—and therefore quite unlike the case of actual spacing in Mongolian or Manchu—, as it is not rare for such alternations to occur within words, and likewise common for word boundaries to fall in between two characters of the same type.

96

sven osterkamp

installment in the series to contain a parallel text in Korean and reading glosses in han’gŭl. It is however the Japanese main text to which small circles are attached here, as expected in the same position as the short strokes in Han Ch’ŏngmun’gam and Pangŏn yusŏk. The principles underlying the use of these circles are not always as straightforward as the mechanical replacement of white spaces with strokes in the case of Manchu (see Tsuji 1991 [1997: 173–192] for an analysis of their usage), but in any case it seems not unlikely that this kind of segmentation for a language that is usually reduced to writing without any spacing is nothing but an extension of what was fijirst introduced in the teaching materials for Manchu and Mongolian. While the Manchu portions of Tongmun yuhae (1748) take over the spacing from Manchu usage intact—despite the use of han’gŭl throughout—the latter works Pangŏn yusŏk (1778) and Han Ch’ŏngmun’gam (1779?) retain the idea of word spacing but use short strokes instead. Interestingly this also applies to the Korean texts of some works printed during the second half of the 18th century, as noted by Tsuji (1997: 192, note 20). It thus appears hardly coincidental that at about the same time a new edition of the Ch’ŏphae sinŏ is published that transfers the idea of segmentation to Japanese. For didactic purposes this was undoubtedly a huge leap forward, but at the same time it also allows us to observe the Korean compilers’ analytical view on Japanese as an almost isomorphic foreign language. Finally to come back to the above-mentioned Ahakp’yŏn once again, where we also fijind cases of a short stroke next to the upper right corner of syllable blocks, now however to mark long vowels in Korean, Japanese as well as English. See table 11 for examples. Type B1: Diacritics taken from Mongolian and Manchu letters A case leaving no doubt that neither Chinese influence was involved nor a genuinely Korean invention is the horn-like diacritical mark employed in the teaching materials for syllable-initial /l/ in Manchu and Mongolian (whereas syllable-fijinal /l/ is indicated by a circle and /r/ left unmarked). It was introduced to cope with the fact that the two languages have a phonemic contrast between /r/ and /l/ (which is also reflected in writing by virtue of diffferent letters) while no such contrast exists in Korean phonology. The modern language merely features an allophonic variation between [l] and a flap [ɾ], and there is reason to believe that the realization as a flap was originally—and still so in Middle Korean when the

early adaptations of the korean script

97

Table 11.  to mark vowel length in Ahakp’yŏn character

Sino-Korean

Sino-Japanese

English

shī 師 zhòng 仲

sa 사 、 cuːng 중

si 시 (=、 shi) cyuː쥬 (= chū)

Teacher Second

transcr. of English 、 thiː-chwə 티 춰 ssɛ-khon-tɨ 쌔콘드

script was created—used in most environments showing [l] today. In any case, Korean orthography is strictly phonemic here and accordingly uses the same single letter irrespective of its actual realization. The distinction between narrow versus broad transcription in the works on Manchu has already been mentioned above. What triggered the use of a narrow transcription was the absence of the original script, in this case the absence of Manchu script. All of the textbooks mentioned in the introduction have the main text written in original script, to which a transcription in han’gŭl as well as a translation into Korean is added. As /l/ and /r/ have separate letters in Manchu script and the text in exactly this script was present, there was no necessity to indicate which of the two was meant in the transcriptions, which were identical for both. The dictionaries are diffferent however: As was the case with the other yuhae for Mongolian and Japanese, the main text of the Chinese–Korean–Manchu dictionary Tongmun yuhae is entirely in Chinese characters and han’gŭl. The Manchu translations of the head words are merely transcribed, but not given in original script. The case of Han Ch’ŏngmun’gam is somewhat diffferent, as at least the translations of the headwords are also found in original script here. The longer defijinitions in Manchu however are in transcription only and accordingly make consistent use of diacritics for /l/ versus /r/—while the transcriptions for the headwords do not. Where did the compilers get their inspiration for the horn-like diacritic from then? Even a cursory comparison of the examples from the Tongmun yuhae (facsimile edition YTTY 1956) given in table 12 with the various latter shapes of in Manchu usage as shown in table 13 immediately reveals its origin: If the “spine” added in gray in the latter table is ignored together with the remaining elements all three letters have in common, it is a horn-like element pointing upwards that remains as the distinctive shape of . The diacritic used in the above-mentioned works is then nothing else but this distinctive element of the letter .10

10  For a work on Manchu outside the tradition of the Bureau see for instance the mid18th century Ch’ŏnghagŭm 清學音 studied in Lee (1973). Here neither special letters nor

98

sven osterkamp Table 12. Diacritics to indicate Manchu /l/ in Tongmun yuhae galaka

ᡤ᠇ᠯᠠᠡᠡ

qíngle 晴了

‘(the weather) cleared up’ (I/3a) bolori

ᠪᠤᠯᠣᡵᡳ

qiū 秋

‘autumn’ (I/3a)

talkiyan

ᠲᠠᠯᠭᡳᠶᠡᠨ

diàn 電

‘lightning’ (I/2a)

Type B2: Diacritics taken over intact from Japanese Initially Japanese phonographic writing used separate syllabograms depending on the presence or absence of (prenasalization and) voicing. Before long the distinction was largely lost however and phonograms formerly used to render syllables with voiceless initials could be used to render those with voiced ones as well. A new means to diffferentiate between the two categories—which remained intact in the language, still forming a phonological opposition—was developed in a situation of language contact: Sinological studies demanded a precise notation of the various classes of initials in Chinese, including the already mentioned one termed (quán)zhuó (Japanese daku), i.e. that of voiced unaspirated obstruents. Tone dots used to mark the tone a given character is to be read in did not only give birth to the so-called side dots in Middle Korean texts. It also led to a parallel tradition in Japan.

diacritics are used in what are evidently transcriptions by ear. Generally both Manchu /r/ and /l/ are rendered by (i.e. /l/ [ɾ]), the latter however tends to be transcribed as (writing, as was usual at that time, Korean /ll/ [lː]) intervocalically. The same strategy of singleton /l/ [ɾ] versus double /ll/ [lː] to render foreign /r/ and /l/ reprectively is still seen in modern loanwords, at least in intervocalic position. In Ahakp’yŏn however this is also extended to English /r, l/ in word-initial position, simply by transposing these initial consonants into intervocalic ones through prosthesis. Cf. liver as 을리 versus river 으리 (phonetically [ɨlːivə] versus [ɨɾivə]; pp. 5, 8).

early adaptations of the korean script

99

Table 13. Letter forms of in Manchu script initial

medial

fijinal





As initial types had influence on tonal developments the diffferent classes of initials were indicated by distinct signs, e.g. a single dot for quánqīng ones, but two dots for quánzhuó ones. Despite their name these tone dots existed in a variety of shapes, including besides dots also circles or short strokes, and it is a double stroke used in this manner that developed into the standard diacritical mark to indicate voiced obstruent initials (called dakuten, or ‘dots [to indicate] zhuó[-type initials]’). Thus for instance タ versus ダ. Now Korean lacks phonemic voicing contrasts. Stops and afffricates are usually voiceless in initial position and only become voiced in voiced environments. While older works on Japanese at the Bureau concentrated on prenasalization rather than the accompanying voicing and introduced new consonant clusters of nasals plus homorganic obstruents as we have seen above, this became less and less of an option with prenasalization being gradually lost in Japanese. It is likely a consequence of this phonetic change that more recent works have at times simply borrowed dakuten marks and applied them to han’gŭl blocks. Probably among the earliest examples is a manuscript closely related to the dictionary Waeŏ yuhae mentioned above, entitled Hwaŏ yuhae 和語類解 (1837),11 but the same can still be observed in the early twentieth century. Table 14 reproduces part

11

 Manuscript in the possession of the library of the Graduate School of Letters, Kyōto University (shelf-mark Philology|2D|41c). For a facsimile edition see e.g. Chung (1988). Note that especially in the second volume the dakuten tend to be placed to the left of syllable blocks.

100

sven osterkamp Table 14. Han’gŭl with dakuten

g-

z-

d-

b-

-a

-i

-u

-e

-o

ガ

ギ

グ

ゲ

ゴ

가゛

기゛ 구゛

계゛

고゛

ザ

ジ

ズ

ゼ

ゾ

사゛

시゛

수゛

셰゛

소゛

ダ

ヂ

ヅ

デ

ド

다゛

디゛

두゛

뎨゛

도゛

バ

ビ

ブ

ベ

ボ

바゛

비゛

부゛

볘゛

보゛

of a syllable chart found on pages 120f. of Nik-Kan Kan-Nichi gengoshū 日韓韓日言語集, published in Tōkyō in 1910 (wrongly assigned glosses were moved to their proper position). The extent however to which these diacritics are actually used in this textbook is nothing compared to Hwaŏ yuhae, or also the horn-like diacritic used for Manchu and Mongolian. Assimilation in Terms of Writing Direction Another interesting aspect of Korean writing used to render foreign languages has less to do with adaptation than with assimilation, namely in terms of writing direction. Following the Chinese model, Korean like Japanese was traditionally and to some extent still is written in columns from

Even if only in passing, it seems noteworthy to mention a rather similar approach in one of the two romanizations of Japanese advocated by Italian orientalist Antelmo Severini (1827–1909). He proposed to employ either a strict transliteration thruthful to the original spelling, with diacritics such as dakuten carried over without change, or a pronunciationbased transcription, depending on the audience and purpose (see Congrès international des Orientalistes 1874: 114–130, 390–417; especially pp. 115–117). The former kind of romanization is found in several works by Severini and other scholars, such as his student Carlo Puini (1839–1924). To name but one example: already in the title of their collaborative publication Repertorio Sinico-Giapponese (Florence 1875) the title of the dictionary Wa-Kan onshaku shogen jikō setsuyōshū 和漢音釋書言字考節用集 is romanized as “Wa kan won seki siyo ゛ken ゛si kau setu you siu”—which not only strictly follows the then-current kana orthography (with e.g. かう for kō or しう for shū), but also retains the dakuten in げん for gen and じ for ji.

early adaptations of the korean script

101

right to left. It is thus only natural to see this direction retained in all premodern Korean works on Chinese and Japanese. Mongolian and Manchu are diffferent however: While they are similarly written in columns, the direction is from left to right instead. At least in part of the materials on Mongolian and Manchu one fijinds the Korean portions to imitate the direction of the surrounding text, they thus proceed from left to right, as would otherwise be quite unusual. To take the works on Manchu as an example, such an assimilated direction of writing is found in all textbooks proper, i.e. Soaron, P’alsea, Samyŏk ch’onghae and Ch’ŏngŏ nogŏltae, namely in all phrase-for-phrase translations of the original text into Korean. The two dictionaries Tongmun yuhae and Han Ch’ŏngmun’gam are under the influence of Chinese, as they are in fact all trilingual (Chinese– Korean–Manchu) with Chinese on top of the hierarchy. As is the case with all works on Chinese alone, the Chinese direction of writing is retained despite the primarily Manchu-oriented content. Interestingly here even the defijinitions in Manchu, which as was noted above are only given in transcribed form here, are written from right to left. As assimilation therefore occurs in both directions it is apparent that the direction of writing for all languages present in a given work is here simply determined by that of the main text or, in the case of the dictionaries, by that of the language used for the head words. References Abel-Rémusat, Jean-Pierre 1820. Recherches sur les langues Tartares, ou mémoires sur diffférens points de la grammaire et de la littérature des Mandchous, des Mongols, des Ouigours et des Tibetains. Tome I.er. Paris: Imprimerie Royale. Bonaparte, Roland 1895. Documents de l’époque Mongole des XIIIe et XIVe siècles. Paris: Imprimé pour l’auteur. Chung Kwang [Chŏng Kwang] 鄭光 (ed.) 1988. Chebon chipsŏng Waeŏ yuhae 諸本集成倭 語類解 [Waeŏ yuhae, a collection of the diffferent editions]. Seoul: T’aehaksa. Clauson, Gerard 1959. The ẖP’ags-pa alphabet. Bulletin of the School of Oriental and African Studies, University of London 22: 300–323. Congrès international des Orientalistes (ed.) 1874. Congrès international des Orientalistes. Compte-rendu de la première session Paris—1873. Tome premier. Paris: Maisonneuve et Cie. Junast [Zhàonàsītú] 照那斯圖 & Yáng Nàisī 楊耐思 1984. Bāsībā zì yánjiū 八思巴字硏 究 [Studies in ‘Phagspa script]. In: Zhōngguó mínzú gŭwénzì yánjiūhuì 中國民族古 文字硏究會 (ed.), Zhōngguó mínzú gǔwénzì yánjiū 中國民族古文字硏究 [Studies in the ancient scripts of China’s minority peoples], Běijīng: Zhōngguó shèhuì kēxué, pp. 374–392. Hattori Shirō 服部四郎 1979. Nihon sogo-ni tsuite 13 日本祖語について・13 [On ProtoJapanese, part 13]. Gekkan gengo 月刊言語 8(3): 87–97.

102

sven osterkamp

Higuchi Kōichi 樋口康一 2001. Pasupa moji パスパ文字 [The ‘Phagspa script]. In Kōno Rokurō 河野六郎, Chino Eiichi 千野栄一, Nishida Tatsuo 西田龍雄 (eds.), Sekai moji jiten 世界文字辞典 [Encyclopedia of the world’s writing systems], Tōkyō: Sanseidō, pp. 727–734. Hope, Earl R. 1953. Karlgren’s glottal stop initial in Ancient Chinese, with particular reference to the hPhags-pa alphabet and to certain points of linguistic psychology. Ottawa: s.n. —— 1957. Letter shapes in Korean Önmun and Mongol hPhagspa alphabets. Oriens 10(1): 150–159. Ledyard, Gari K. 1966. The Korean Language Reform of 1446. Doctoral dissertation, University of California, Berkeley. —— 1998. The Korean Language Reform of 1446 (Kungnip kugŏ yŏn’guwŏn ch‘ongsŏ 國立 國語硏究院叢書 2). Seoul: Sin’gu munhwasa. Lee Ki-Moon [Yi Kimun] 李基文 1973. 18 segi-ŭi Manjuŏ pangŏn charyo 十八世紀의滿 洲語方言資料. [A source on an 18th century dialect of Manchu]. Chindan hakpo 震 檀學報 36: 99–132. Meister, Georg 1692. Der Orientalisch-Indianische Kunst- und Lust-Gärtner. Dresden: Johann Riedel. Müller, Andreas 1694. Syllabarium Japanicum geminum. E Manuscripto Meakensi alterum; alterum E Sinarum Traditione. Alphabeta Universi. Königsberg: Gottfried Bartsch. Nam Kwangu 南廣祐 1997. Koŏ sajŏn 古語辭典 [Dictionary of pre-modern Korean]. Seoul: Kyohaksa. Overmeer Fisscher, J.F. van 1833. Bijdrage tot de kennis van het Japansche rijk. Amsterdam: J. Müller & Comp. Siebold, Philipp Franz von 1832–[1857?]. Nippon. Archiv zur Beschreibung von Japan und dessen Neben- und Schutzländern. Leiden: Bei dem Verfasser. Tawata Shin’ichirō 多和田真一郎 1979. 15, 16 seiki Shurigo-no on’in (ge) 十五・十六世 紀首里語の音韻 (下) [The phonology of 15th to 16th century Shuri Ryūkyūan, part 2]. Okinawa bunka 沖縄文化 52: 23–42. Tsuji Seiji 辻星児 1991. Jūkan kaishū Shōkai shingo-ni mirareru kugiri shōten-ni tsuite 重 刊改修捷解新語に見られる区切り小点について [On the punctuation marks in form of small circles seen in Chunggan kaesu Ch’ŏphae sinŏ]. Jisho, gaikoku shiryō-ni yoru Nihongo kenkyū 辞書・外国資料による日本語研究 [Studies in Japanese language based on dictionaries and foreign sources], Ōsaka: Izumi shoin, pp. 395–418. —— 1997. Chōsengoshi-ni okeru Shōkai shingo 朝鮮語史における『捷解新語』 [Ch’ŏphae sinŏ in the history of the Korean language] (Okayama daigaku bungakubu kenkyū sōsho 岡山大学文学部研究叢書 16). Okayama: Okayama daigaku bungakubu. Wall, Charles William 1840. An examination of the ancient orthography of the Jews, and of the original state of the text of the Hebrew bible. Vol. II. London: Whittaker and Co. Wylie, Alexander 1870. On an Ancient Buddhist Inscription at Keu-yung-kwan, in North China. The Journal of the Royal Asiatic Society of Great Britain & Ireland, New Series 5 (1): 14–44. YTTY = Yŏnhŭi taehakkyo tongbanghak yŏn’guso 延禧大學校東方學硏究所 (ed.) 1956. P’alsea, Soaron, Samyŏk ch’onghae, Tongmun yuhae 八歲兒 小兒論 三譯總解 同文類 解 (Kukko ch’onggan 國故叢刊 9). Seoul: Yŏnhŭi taehakkyo tongbanghak yŏn’guso. Yule, Henry (tr.) 1871. The book of Ser Marco Polo, the Venetian. Concerning the Kingdoms and Marvels of the East. 2 vols. London: John Murray.

HAN’GŬL REFORM MOVEMENT IN THE TWENTIETH CENTURY ROMAN PRESSURE ON KOREAN WRITING Thorsten Traulsen Introduction The national context: In 1894 Korea started the far-reaching “reforms of the year kabo” (kabo hyŏngmyŏng) under Japanese aegis. The Shimonoseki peace treaty after Japan’s victory over China (1895) marked the end of Chinese influence on the peninsula. At the core of these reforms was the promulgation of national sovereignty and the decentering of China. The new model for the political agenda was the Western style nation state. Within the following decade, Korea was to undergo a thoroughgoing modernization in almost every single aspect of everyday life with Japan serving as a role model as it had been in the 1880s when Korean reformers fijirst began to study there. Most important of the governmental activities for the linguistic context of this article is the following passage from the Kabo edicts (quoted from King, 1998: 37): “All laws and edicts should have kungmun [i.e., pure Korean] as their base; one may attach a translation in hanmun [i.e., written Chinese] or mix kungmun and hanmun together.” The last years of the nineteenth century also mark the beginning of a newly emerging scholarly interest in Korean language and writing system going hand in hand with the construction of the image of a nation (one people—one language—one nation). Activities like proposing the “unity of written and spoken language” (ŏnmun iltch’i)1 to reach a “national text (style)” (kungmun), i.e., to overcome the diglossic situation of written Chinese and spoken Korean (as well as written hybrid-forms like idu) that dominated the last at least 1,500 years, led to the writing of Korean grammars, discussions of orthographic principles to unify and update the confused historical spelling, editions of Western style school books, etc. One of the most influential publications of these days was the Korean

1

 This idea was vividly discussed in Japan since the 1870s together with kana-only proposals (i.e., the abolition of Chinese characters for Sino-Japanese words) as well as proposals for the adoption of Roman characters. The corresponding Sino-Japanese term gembun’itchi 言文一致 was coined by Kanda Kōhei in 1885 (Nanette Twine 1991: 293).

104

thorsten traulsen

newspaper “The Independent” (Tongnip sinmun, 1896–1899) written in the Korean han’gŭl alphabet only. It was conceived and published by Sŏ Chae-p’il (1864–1951, aka Philip Jaeyson), a Korean reformer who, after the failed Kapsin coup d’etat in 1884, in which he took part, went to Japan and America. In 1896 he came back to Korea where he hired the young Chu Si-gyŏng (1876–1914, aka Han Himsaem) for his newspaper project. While Sŏ Chae-p’il was expelled from the country in 1898 and did not come back before 1947, Chu Si-gyŏng was to become the “father” of modern Korean linguistics and is still treated as a national hero. The Independent became famous for introducing features like word spacing (pin k’an ttŭiŏ-ssŭgi) and han’gŭl-only into Korean writing.2 It served both implicitly as a model for pure Korean writing as well as explicitly as a medium for discussing linguistic matters in its editorials. Beside the pragmatic dimension of these activities toward a modern nation state, there was the linguistic-nationalistic dimension of positioning the Korean language and han’gŭl in a new world order. While a purely linguistic arguing in grammatical terms would have meant to prove that Korean is a prestigious flectional language like Latin, Greek or Sanskrit (and not an agglutinating as it is), it was the Korean alphabet that served best for a national-linguistic self-esteem. The reasoning was as follows: there are “sound-scripts” (sori kŭl) or—as Chu Si-gyŏng wrote in his fijirst editorial for The Independent (April 22 and 24, 1897, cf. King 1998: 43) sketching the history of writing—“true writing systems” (ch’am kŭltcha) like the Roman letters (romaja) and han’gŭl; and there are meaningscripts (ttŭt kŭl)—or “pictures” (kŭrim), i.e., not writing—like Chinese and Egyptian. The strongest and technologically most advanced countries in world have alphabets, while states with meaning-scripts are backward (China) or extinct (Egypt). The superiority of sound-writing is, according to contemporary scholars, not only proven by the distribution of power in the world, but can also be argued for based on pragmatic reasons: the number of characters is much less than in meaning-scripts, so its easier to learn, easier to read, easier to write, and easier to print.3

2  Folk knowledge attributes these features to Chu Si-gyŏng, but they were already used in the inaugural issue (April, 7 1896) when Chu Si-gyŏng was not yet part of the board of editors (cf. King 1998: 42). 3  A third line of a more implicit kind emphasized the old theory that Han’gŭl is derived from the Indian Devanagari script and thus Korean can be related to the most prestigious language in Western understanding, namely Sanskrit (e.g., Chung Kei Won 1938, An Chasan 1928).

han’gŭl reform movement in the twentieth century

105

Fig. 1. Two examples for compounding syllable blocks from individual letters in han’gŭl.

What makes the Korean alphabet even more prestigious and the “most splendid alphabet in the world” (kajang usu-han kŭl), to use a contemporary phrase, is its historically transparent “sophisticated grammatogeny” (cf. Daniels 1992: 85–86) by King Sejong and his scholars based on Chinese medieval phonological theory and the systematic treatment of phonological features as well as the iconic representation of the place of articulation. However, there was one obstacle to this view: the fact that Korean writing, although being a true alphabet, is not segmental-linear but syllabic-compound. While the number of individual characters is low, the number of possible syllable compounds needed in printing is bringing Korean writing back into the cumbersome realm of logographic writing (fijig. 1). From a technical point of view there are two solutions to reach the efffijiciency of a segmental-linear alphabet: (i) writing Korean in a segmentallinear manner; (ii) adopting Roman letters. While the latter was never seriously discussed (in contrast with Japan, where it was a regular item on the language reform agenda; see note 1), the former was attracting ongoing attention among linguists during the Colonial period (1910–1945) as will be shown in part I. Although this aspect of the han’gŭl movement4 never reached any popularity outside a small circles of linguistically interested nationalists, it left subtle traces in the second half of the twentieth century, traces that will bring us back the very beginning of han’gŭl nationalism, i.e., the fijirst han’gŭl-only newspaper The Independent (Part II). The arguments brought forward in favour of segmental-horizontal-leftto-right writing (hereafter: online writing) were basically the same as for sound-writing: (i) easier to learn, because the number of characters is

4  “Han’gŭl Movement” (han’gŭl undong) is a cover term for all nationalistic activities in the fijirst half of the twentieth century concerning Korean language and writing. These activities are promulgating han’gŭl-only writing and discarding Chinese characters even for Sino-Korean words, coining native Korean words for Sino-Korean vocabulary, writing grammars and dictionaries, as well as promulgating han’gŭl online writing (at least by some of the main fijigures).

106

thorsten traulsen

small (ii) easier to read, because eye movement is more natural horizontally than vertically (iii) easier to write, because in left-to-right writing the writing hand of (outnumbering) right-handers is neither hiding nor smudging what has been written before; (iv) easier to print, because less type sorts are needed. A fijifth argument hardly explicitly mentioned, but at least in one case put into practice is (v) the linear sorting of words in dictionaries, indices, etc.. In syllabic compound writing it is syllables that are sorted, e.g., although having the fijirst three segments in common, the word /ka.mul/ would not be listed next or close to /kam/ since the fijirst syllable is diffferent. In 1936 the Korean Language Society (Chosŏnŏ hakhoe) used online writing (beside syllabic compound writing) for the index of its “Revised standard dictionary of Korean” (Sajŏng-han Chosŏnŏ p’yojunŏ moŭm). This set of “pragmatic” arguments was sometimes expanded by a more “scientifijic” reasoning based on iconicity: linear writing is following the (presumed) linear order of speech sounds (e.g., Kim Tu-bong 1922:B7–8). Iconicity is in fact one of the han’gŭl features that is explicitly stated in its birth document (Hunmin chŏngŭm, 1443/4). The basic consonant graphemes are “formed according to the shape” of the corresponding place of articulation. In contemporary discussions iconic means natural means scientifijic means superior.5 Part I: Ideas & Proposals While the actual problems of syllabic writing could be overcome rather easily by simply writing han’gŭl letters online, proposals building upon Chu Si-gyŏng after his death in 1914 went far beyond that. His disciples did not only come up with ideas of how to deal with the technical aspects but pushed the discussion into a direction of an imitation of Western writing that made Korean look like a mixed Roman-Cyrillic-character text, but without being a transcription (i.e., adoption of Roman letters). The purely technical problems can be listed as follows and are partly overlapping with contemporary discussions of orthography:

5  It is interesting to see how pride is taken in the fact the fact that Han’gŭl is harmonizing extremes like phonetic and iconic writing (the latter usually associated with semantography) as well as featural and syllabic writing (instead of being simply segmental, e.g., Kim, Chin W. 1997: 159).

han’gŭl reform movement in the twentieth century

107

i. Modifijication of letter forms, in particular the three graphically primitiv letters /â/ (a single dot), /ŭ/ (horizontal stroke), /i/ (vertical stroke). These letters where right from the beginning of online writing considered to be too minimal and have been omitted (/â/, s. below), severely reshaped (/ŭ/), or in the case of the vertical stroke /i/ (reminiscent of its Roman counterpart) slightly modifijied by serifs to /I/. ii. The treatment of graphemic double or triple vowels (the pronunciation of these diphthongs has already undergone monophthongization): writing them together as one grapheme or as a sequence of two single vowel graphemes. There are two types of graphemic diphthongs: a. offfglides /Vi/ written as simple vowel + /i/ (“separate /i/” ttan i) in one syllable block (fijig. 2).

Fig. 2. Graphemic composition of offfglides /Vi/ in han’gŭl.

b. onglides /wV/ as in /wa/ or /wŏ/ written as round vowel /o/ or /u/ for the labial glide plus the main vowel /a/ or /ŏ/ respectively in one syllable block (fijig. 3).

Fig. 3. Graphemic composition of labial onglides /wV/ in han’gŭl.

The third type, the palatal onglide /yV/ as in /ya/, /yŏ/, /yo/, /yu/, is built not by composition of two simple vowels but by stroke addition, thus cannot be decomposed straightforwardly (fijig. 4).

Fig. 4. Graphemic derivation of palatal onglides /yV/ in han’gŭl.

108

thorsten traulsen

iii. The treatment of the single-dot letter “deep-a” /â/: the pronunciation of this letter was not distinctive anymore and had merged with either /a/ (initial syllables) or /ŭ/ (non-initial syllables). Contemporary orthographic discussions made diffferent proposals, e.g., using /a/ in initial and /â/ in non-initial syllables. The online writing movement simply ignored this letter, followed the pronunciation, and used /a/ or /ŭ/ instead. iv. Directly linked to syllabic compound writing is the double usage of ‘O’ (circle) for /ng/ in syllable fijinal position and ‘zero’ in initial position (= vowel initial). In online writing it is not possible to keep these two functions apart, but the latter is not needed anyway. All online-writing proposals simply omit the zero-initial usage. v. The treatment of tensed consonants: they were traditionally written as ‘sC’-clusters with an initial ‘s’-letter indicating tensed pronunciation of the following consonant /C/. However, all online writing proposals took the modern view and standard since 1933, i.e., writing tensed consonants as double-consonants (/kk/, /tt/, /pp/, etc.). Proposals only difffer in whether to write these double consonants as two separate instances of a simple consonant or come up with a ligature. vi. Morpheme- or word-spacing: while traditionally there was no spacing in Korean writing, this principle would make online writing without any visual clues in form of syllable compounds almost impossible to read. All segmental-writing proposals include spacing. vii. Writing left-to-right or right-to-left, top-down or bottom-up: all online writing proposals follow the Western tradition of writing left-toright, top-down. Additionally, the problem of whether and how to indicate distinctive vowel length was touched upon in some proposals but this will not be dealt with here. While all these topics did not rise any serious problems, some proposals of online writing additionally tried to ape Roman characters by imitating features associated with it but not necessary for the solution of the problems mentioned above. These are: (i) minuscule vs. majuscule letter forms—and how to use them: majuscule for the initial of proper nouns and sentences (the English way, Kim Tu-bong 1922); minuscule for sufffijixes (mach’in poram 1913, s. below); majuscule for (nucleus) vowels (Yi Kyu-yŏng 1911–13); (ii) cursive style for handwriting; in fact, the most far-reaching proposals have been made in the context of cursive writing; (iii) serifs: not discussed, but put into practice; (iv) punctuation (will not be dealt with here). What follows is an exposition of the early and most important proposals for online writing.

han’gŭl reform movement in the twentieth century

109

Chu Si-gyŏng (1908, 1909, 1913, 1914): While in his fijirst editorial for The Independent (1897, s. above) the young Chu Si-gyŏng claimed the superiority of han’gŭl due to its syllabic compound writing (cf. King 1998: 43)—presumely because it added a further level of linguistic analysis to writing—, the beginnings of horizontal-segmental writing can be traced back to him and his activities in the National Script Research Institute (Kungmun Yŏn’guso). In the “Korean national script research draft” (Kungmun Yŏn’gu an) from 1908 he wrote. “It might seem appropriate in orthography to write online [sino-korean. hoengsŏ] each of the consonants and vowels, but of course we should follow the traditional custom and depend upon the precedent of the Hunmin chŏng’ŭm, the original promulgation document of the han’gŭl (1443/4). It is my view that it might be possible to write the letters on-line only in cases where it is necessary.” (quoted from King 1997: 222). The next year, however, in his comment on resolution 11 (on horizontal-segmental writing) of the “Report of the National Script Research Institute” (Kungmun Yŏn’guso chŏngbosŏ), he was more explicit: “In the matter of orthography, we can follow the example of the Hunmin chŏng’ŭm and rely upon the traditional method already worked out. But it is also reasonable to write the consonants and vowels separately online, whether vertically or horizontally . . . This is smoother for the flow of pronunciation, is also convenient pedagogically, and is important for type casting. For example, “Our country is bright and beautiful (uri nara ka palkko kopta).” would be: . . . ” (id. 223).

Fig. 5. Chu Si-gyŏng’ fijirst example (1909) of how to write online (from Kim Min-su 1984: 255).

There are some other early instances of online writing connected with the activities of Chu Si-gyŏng (King 1997: 223, Kim Min-su 1984: 144–145, Kim Chŏng-su 1982) like a series of features in the monthly journal “Children’s pictorial” (Ai-dŭl poi, 1913.09–1914.10). Or the “certifijicate of completion” (mach’in poram) issued by the Korean Language Society (Chosŏn Ŏnmunhoe) in 1913/03/02. In the latter case, sufffijixes are written in halfsized letters and the han’gŭl character /ŭ/ (horizontal stroke) is reduced

110

thorsten traulsen

to a dot. All these early uses hint at early 1913 as the time where online writing became popular as a graphical icon for the han’gŭl-movement.

Fig. 6. Chu Si-gyŏng’s ‘Certifijicate of completition’ (Mach’in poram, 1913, left) and ‘The Children’s Pictorial’ (Ai-dŭl poi, 1914.03, page 45, right) (from Kim Min-su 1984: 255).

Chu Si-gyŏng’s attitude toward online writing was fijinally summerized in his “Practice of writing our script horizontally” (Uri kŭl-ŭi karo-ssŭnŭn ikhim, this time using the native Korean term karo-ssŭgi instead of sinokorean hoengsŏ, both meaning “horizontal writing”) on the last page of his posthumous opus “The speech sounds” (Mal-ŭi sori, 1914): That which is in accord with Heaven succeeds. That which contravenes Heaven fails. If one wishes to live as well as possible, one must make things that way. Living life as well as possible lies in following Heaven. On this earth, each people has, since time immemorial, its own sounds, and these are their languages. This is Heaven. In creating the best life possible, man needs writing. For each people to create and use a writing system appropriate to its language is to follow Heaven. To create good language, one must cultivate language; good writing comes about through writing good language. The best writing is that which writes the most-cultivated language, and which writes segmentally left to right (karo ssŭnun kŏs ira). Segmental, left-to-right writing (karo kŭl) is best for writing, reading, and printing. L T M C [i.e., 4357] one winter day Han Himsaem [i.e., Chu Si-gyŏng] (translation apart from the last line by King 1998: 49)

han’gŭl reform movement in the twentieth century

111

This text not only states the necessity of writing on-line, but is actually written in this fashion:

Fig. 7. Chu Si-gyŏng pleed for online writing (from Mal ŭi sori, 1914, left). Right side: the same text rendered in syllabic compound writing.

Chu Si-gyŏng modifijied the letter /ŭ/ (horizontal stroke) to “H” (not seen in the examples before) and added serifs to the /i/ (vertical stroke) → “I”. Apart from “H” all letters are easily recognizable. The spacing is between morphemes, not words. Diphthongs and tensed consonants are treated as two individual letters. Another feature of Roman letters used here at the end of the text is the use of letters as numbers (here: according to their position in the Korean alphabet) to indicate the year, in this case ‘L T M C’ = 4357 (however,

112

thorsten traulsen

one would expected ‘L N L C’ for the year Tan’gi 4247 = 1914 as in other examples of the same year).6 Koreans in Russia (1913/14, 1925): The monthly “Journal of the Orthodox Koreans in Russia” (Taehanin chŏnggyobo) was published in eleven issues during 1913 and 1914. Volumes 9 to 11 contain texts written online (for reproductions, translations and analyses see King 1997: 225–239). The letter shape of /ŭ/ (horizontal stroke) has been changed to “U”, a practice that can be found again in Kim Tu-bong’s proposal a few years later (s. below). A later example of online writing in Russia from 1925 can be found in the newspaper “The vanguard” (Sŏnbong), again using ‘U’ for /ŭ/ (for a reproduction see Pak Hwan 2008: 90). Yi Kyu-yŏng (1911–13): Yi Kyu-yŏng (1890–1920) was one of Chu Si-gyŏng disciples. The handwritten records of his researches “All kind of things” (Ongat kŏt, 1911–13) are included in the “Collected works of Chu Si-gyŏng” (Chu Si-gyŏng chŏnsŏ). Ten pages (486–493) of notes on Korean morphology and parts of speech are written online. The system is very much in the manner of Chu Si-gyŏng (1914), i.e., the vowel /ŭ/ (horizontal stroke) is written as “H”, /i/ (vertical stroke) as “I”, as well as showing morpheme and not word spacing. What is new in these manuscripts is the use of consonants and vowels in diffferent sizes: consonants are about 1/3 smaller than vowels, offfglides (Vi) are rendered as “small V + big /i/”:

Fig. 8. The fijirst line in online writing from Yi Kyu-yŏng’s ‘All kind of things’ (Onkat kŏt, 1911–13, from Chu Si-gyŏng chŏnsŏ, vol. 4, p. 486). In syllabic writing this line would look like:

6  Tan’gi is the abbreviated form of Tan’gun kiwŏn (“Tan’gun origin”) with Tan’gun being the mythological founder of Korea. Tan’gi 1 corresponds to 2,333 BC. So far, I was not able to fijind a alternative calculation making Chi Si-gyŏng’s usage of ‘L T M C’ meaningful.

han’gŭl reform movement in the twentieth century

113

This practice of “majuscule vowels” can be found again on another “Certifijicate of completion” (mach’in poram) issued by the Korean Language Society (Chosŏn Ŏnmunhoe) in 1915. Kim Chŏng-su (1982: 16) attributes this to another disciple of Chu Si-gyŏng, Kim Tu-bong (s. below), but his use of majuscule and minuscule is diffferent as we will see. This time the year given in the han’gŭl letters ‘L N L NG/ and can easily be read as (Tan’gi) 4248 = 1915.

Fig. 9. A Certifijication of completion (Mach’in poram, 1915.03.31) (from Kim Chŏng-su 1982: 15).

Kim Tu-bong’s (1922): Kim Tu-bong (1889–1961?), one of Chu Si-gyŏng’s most faithful disciples, added to the “Amended and extended” (kiptŏ) reedition (1922) of his “Korean grammar” (Chosŏn malbon, 1916) a lengthy treatise of 102 pages on “good writing” (chohŭl kŭl, i.e., online writing), “stenography” (nalchyŏk), and “standard language” (p’yojunmal). Apart from replacing Chu Si-gyŏng’s “H” for /ŭ/ (horizontal stroke) by a curved “U”, introducing punctuation as well as non-obligatory prosodic and repetition marks, his system and his letter shapes are as straightforward and easily recognizable as Korean as in the systems before. Furthermore, he mentions the possibility of writing the “fijirst letter of proper nouns” (hollo im-ŭi ch’ŏt kŭlssi) bigger than the others (1922:B16). This practice can be found on the cover page for the title of the included CHohŬlKŭl (“good writing”), the place of publishing SangHai (“Shanghai”, s. fijig. 10), and the name of the publishing house SaiKŭlCHip (“New Letter House”, s. fijig. 10):

114

thorsten traulsen

Fig. 10. Online writing with capital letters from the cover page of Kim Tu-bong’s Kiptŏ Chosŏn malbon 1922.

Furthermore, Kim is giving new letter shapes for writing in a “floating style” (hŭllimch’e), i.e., cursive handwriting that combines the letters of a word (inlet between page B20 and B21 of the appendix). The letter shapes in isolated form are in most cases still reminiscent of the original han’gŭl letters they replace, but the text sample he is giving is hardly readable or looking like Korean. The main diffference between traditional han’gŭl letters and the new cursive style is ‘roundness’. The feature of “round letter shape” for a “good script” has already been discussed by him when arguing for the replacement of /ŭ/ (horizontal stroke) by ‘U’ (1922:B4–5).

Fig. 11. Cursiv han’gŭl letter shapes according to Kim Tu-bong 1922.

Ri P’il-su (1923, 1925): The most “sophisticated” approach to remodelling Korean han’gŭl letters for horizontal writing was made by another one of Chu Si-gyŏng’s disciple, Ri P’il-su, in his “Grammar of correct sounds” (Chŏngŭm munjŏn, 1923). In fact, in his chapter on “Grass letters of our script” (Uri kŭltcha-ŭi chy’osŏ, i.e., cursive script, pp. 138–148) he went all the way and actually borrowed Roman cursive letters. However, instead of simply transcribing Korean by means of them, he was looking for graphical relations within the Roman cursive letter shapes similar to those of the han’gŭl alphabet, i.e., stroke addition (mainly for aspirates), doubling (for geminates), or mirroring (for vowel quality):

han’gŭl reform movement in the twentieth century

115

Fig. 12. Systematic graphical manipulations in han’gŭl.

He came up with pairs or triplets like ‘ɑ’ → ‘d’ ≡ ‘o’ → ‘p’, ‘n’ → ‘m’ ≡ ‘v’ → ‘w’, ‘i’ → ‘j’ (‘≡’ meaning “showing the same relation as”). Then he assigned one basic han’gŭl grapheme to one of his basic Roman letters (in some cases on the basis of sound identity, e.g., ‘ɑ’ for /a/, so far in fact being a transcription) and derived the letter shapes for the other sounds according to graphical derivation (fijig. 13).

Fig. 13. Some illustrative examples of Ri P’il-su’s proposal of how to use Roman cursive characters to write Korean online (fijirst row: han’gŭl letter; second row: assigned Roman cursive character; third row: sound value).

The assigned Roman minuscule can be capitalized for writing initials. His proposal was re-published in an elaborated form two years later in “Questions and answers about writing” (Kŭl e tae-han mundap, 1925: 61–85) by the “Korean Society for Correct Grammar” (Chosŏn Chŏngmunhoe) in the fijinal chapter on “grass script” (ch’osŏ), i.e., cursive writing. Ch’oe Hyŏn-bae (1922, 1926, 1937, 1947): Hooking into the discussion of cursive letter shapes for han’gŭl, another of Chu Si-gyŏng’s disciple, Ch’oe

116

thorsten traulsen

Hyŏn-bae (1894–1970, pseud. Oesol) presented his proposal in the Chosŏn Ilbo newspaper (1922, 1926), the latter article at the occasion of the fijirst celebration of han’gŭl day (then: ka-kya-nal; today: han’gŭl nal) (cf. Kim Yun-gyŏng, 1937: 14). His idea is similar to—though more legible than— Kim Tu-bong’s in so far as he readily accepts Roman letter forms reminiscent of the intended modifijied and ‘rounded’ han’gŭl letter (fijig 14).

Fig. 14. Proposals for replacing han’gŭl letters with similar looking Roman letters.

Fig. 15. Ch’oe Hyŏn-bae’s proposal for cursive han’gŭl letters (1926) with majuscule (upper block, four lines) and minuscule (lower block, four lines) letter forms. Corresponding printing letters (traditional forms plus three additions glides in online writing) are written beneath the proposed cursive forms. The fijirst two rows of a block displays the consonants and semi-vowels (proposed and traditional); the third and fourth row shows the vowels (id.) (from Kim Chŏng-su 1982: 17).

Glides are represented by separate letters diacritically derived from its associated main vowels: offfglides /Vi/ by a diacritically marked /i/ (fijirst row, last letter, compare with third row, last letter); onglides /wV/ with a diacritically marked /o/ (fijirst row, third from the right, compare with third row fijifth from the left) for the /w/ in /wa/ (in syllabic writing /oa/) or /u/ for the /w/ in /wŏ/ (syllabic: /uŏ/). Much later, in his “revolution of letters” (Kŭlcha-ŭi hyŏngmyŏng, 1947), Ch’oe Hyŏn-bae modifijied his proposal somewhat and modelled a printing style font derived not directly from the angular han’gŭl letters but from the cursive style he invented before. The outcome at fijirst sight looks like a mixed Roman-Cyrillic script (fijig. 16).

han’gŭl reform movement in the twentieth century

117

Fig. 16. Ch’oe Hyŏn-bae’s proposal for majuscule (1) and minuscule (2) letter shapes for printing and handwriting (3 and 4 respectively).

Ch’oe Hyŏn-bae writes that he had the idea for this layout while in prison during the last three years of the colonial period, i.e., 1942–1945 (1947: 133). But his four-fold proposal looks suspiciously like the proposal of Kim Sŏkkon (1932, s. below). Ch’oe Hyŏn-bae was to become the most persistent fijighter for horizontal writing in Korea far beyond the colonial period. As late as 1963 he published a “Reader in Korean online writing” (Han’gŭl karo kŭlssi tokpon). Kim Sŏk-kon (1931, 1932, 1937) wrote some articles on online writing in the Tonga Ilbo newspaper and might have had some influence on Ch’oe

118

thorsten traulsen

Hyŏn-bae (cf. Kim Chŏng-su 1982: 19). He comes up with both a cursive and a printing font each having minuscule and majuscule forms (fijig. 17).

Fig. 17. Kim Sŏk-kon’s proposal for new letter forms (1932). The columns are: (i) number; (ii) majuscule printing font; (iii) minuscule printing font; (iv) majuscule cursive font (v) minuscule cursive font (vi) name of han’gŭl letter (beginning with and ending in it). From Kim Yun-gyŏng (1937: 12–13).

han’gŭl reform movement in the twentieth century

119

Glides are treated separately as in Ch’oe Hyŏn-bae (1926), i.e., by diacritical marking (a dot on the top) of the associated main vowel (fijig. 18).

Fig. 18. Kim Sŏk-kon’s (1932) glides. Columns as in table above.

His proposal is unique in having ligatures for the double consonants (fijig. 19):

Fig. 19. Kim Sŏk-kon’s ligatures for geminates. Columns as in table above.

Korean Language Society (Chosŏnŏ Hakhoe, 1936): It was already three years after the inaugural issue of their journal Han’gŭl (1933) when the Korean Language Society launched the “temporary proposal for writing online” (Karo kŭlssi imsi an), although the inner cover sheet always had the journals title in online writing. Ignoring the discussions of cursive styles as well as majuscule and minuscule forms, their proposal is straightforward with unmodifijied consonants (upper row) and ‘hooked’ as well as partly prolonged vowel shapes, that are still recognizable (lower row):

120

thorsten traulsen

Fig. 20. Letter shapes of the Chosŏnŏ Hakhoe for online writing (1936) (from Kim Yun-gyŏng, 1937: 15).

These letter shapes were used in a series of online reading exercises in Han’gŭl (1938) side by side with the same text in syllabic compound writing, as well as for the index of the “revised Korean Standard dictionary” (Sajŏng-han Chosŏnŏ p’yojunŏ moŭm) project, once again side by side with syllabic compound writing. Some more proposals were made the following year by Kim Pyŏng-ho (1938), Sim Sang-sŏl (1938), Sim Il (1938), all of them building up on existing ideas (cf. Kim Chŏng-su 1982 for illustrations). However, the more and more restrictive language policy of the Japanese administration culminating in arresting the members of the Korean Language Society for compiling a dictionary in 1942 led on the one hand to a temporary end of discussions on online writing. On the other hand this event bestowed reputation on the han’gŭl movement and anchored its members, their activities, as well as the Korean alphabet even deeper in a nationalistic context. Right after the liberation the topic of online writing was put back on the agenda (s. below, part II). Chang Pong-sŏn (?): Another strain of post-war proposals much less in number is called “half-online writing” (pan p’urŏ-ssŭgi). Here only fijinals are written ‘online’, while the (C)V-part of syllables is composed in the traditional manner. Chang Pong-sŏn’s example can not be dated (fijig. 21; Kim Chŏng-su 1982: 23):

Fig. 21. Chang Pong-sŏn’s “half-online writing”.

Yi Hyŏn-bok (1980, 1981) made a similar proposal (fijig. 22; id.:24):

Fig. 22. Yi Hyŏn-bok’s “half-online writing”.

han’gŭl reform movement in the twentieth century

121

Kim Chŏng-su (1982): In an article titled “The development of han’gŭl” (Han’gŭl-ŭi kaebal-e wi-han yŏn’gu, 1982) published in the journal Han’gŭl, Kim Chŏng-su not only outlined the history of online writing so far, he also made a new proposal trying to reconcile the technical advantages of horizontal-resolved writing with the traditional ‘look’ of syllable compound writing by slanting the original han’gŭl letters (including ‘zero’initial 0) by 45° to the left (kiullyŏ-p’urŏ-ssŭgi; fijig. 23; id.:28):

Fig. 23. Kim Chŏng-su’s proposal for “slanted writing”.

Yi Hyŏn-bok (1971, 1981), Kim Chŏng-su (1982): Finally, online writing was the default setting of all proposals for an expanded han’gŭl replacing the international phonetic alphabet (IPA). The fijirst person to sketch out such a system was Yi Hyŏn-bok (1971, 1981; cf. Kim Chŏng-su 1982: 31).7 Once again Kim Chŏng-su was having his own view on the topic and came up with his own proposal for a han’gŭl IPA (1982, the same article that introduced his “slanted writing”, s. above). Summary so far: The impact of Western civilization in general and the Roman alphabet in particular turned the transmitted cultural hierarchies upside-down: phonetic writing, long time sneered at as “women’s writing” (amk’hŭl), “children’s writing” (ahaetkŭl) or more formally as “vernacular writing” (ŏnmun), was now all the rage, at least in the eyes of linguists eager to reform and defijine the “national language” (kungmun) for a modern age. The Korean han’gŭl alphabet, although already an alphabet in 7

 Given the fact that the phonological theory behind the han’gŭl consonants is based on a two-dimensional matrix of place of articulation and manner of articulation, it is interesting to note that no han’gŭl-IPA has been proposed before. In fact, han’gŭl could have been argued for as being a “better” IPA since it does not show the IPA inconsistency of using diffferent characters for voiced and voiceless stops instead of using one character for place of articulation and one diacritic or systematic graphical manipulation for manner of articulation. The original han’gŭl layout of 1443/44 introduced the doubling of voiceless stops to indicate voiced stops needed for a “correct”—this time meaning historically artifijicial—writing of Middle Chinese syllables, but this practice has been given up in the 16th cent. and Sino-Korean was now written with simple stops according to the actual Sino-Korean pronunciation. But even an article in the Han’gŭl journal (editorial board, 1938) discussing the major IPA modifijications of 1932, is not making this point.

122

thorsten traulsen

the closest sense, was subject to reform ideas as well. The impetus and reasoning herein was based on a critical view of the syllabic compound writing of han’gŭl cumbersome for printing and typing, and calling for ”online writing” (hoengsŏ / karo-ssŭgi / p’urŏ-ssŭgi). However, reform proposals went much beyond a simple solution that could have overcome the technical problems and moved more and more towards an imitation of Roman letters but without ever being a transcription. The mainly diachronic process from stimulus difffusion to imitation in online writing can be sketched as follows: i. no influence: the time from the impact of the Roman alphabet in the 1880s to the earliest proposals of online writing. However, the idea of word spacing and punctuation that appeared in The Independent (see above) can be regarded as stimulus difffusion related to writing. ii. stimulus difffusion (core phase 1908–1922): In the beginning, online writing proposals triggered by the contact with the Roman alphabet were straightforward with only slight modifijications in letter shape of those forms considered to be to simple to be distinctive in isolation (i.e., not in syllable compound form). The only Roman feature imitated here almost right from the start was the distinction between minuscule and majuscule. This approach can be found in Chu Si-gyŏng (1908, 1914) and his disciples Yi Kyu-yŏng (1911–13), Kim Tu-bong (1922), but also some time later in the proposal of the Korea Language Society (Chosŏnŏ Hakhoe, 1936). iii. receptive imitation (1922–1943): Although one of the main arguments for online writing was “better for printing”, it was the imitation of Roman cursive hand writing style that made the discussion lift offf. The originally angular letter shapes of han’gŭl were curved, sometimes added with serifs, and—in case there was a candidate—readily replaced by a Roman letter with similar shape. Serifs now play an important role to give letter shapes a visual Roman impression. iv. transmitting imitation (1923/25): While the ‘receptive imitation’ in the above mentioned proposals was uni-directional in so far that features of Roman letters were imitated for han’gŭl letters, in ‘transmitting imitation’ fijirstly graphemic han’gŭl features were looked for in Roman letter shapes, than the Roman letters were borrowed according to their ‘han’gŭl-features’. This is the rarest case of Roman influence on han’gŭl. The locus classicus is Ri P’il-su (1923, 1925). v. adoption: a straightforward adoption of Roman letters (= transcription) was not proposed.

han’gŭl reform movement in the twentieth century

123

From a semiotic point of view the impact of Roman letters on the reform ideas sketched above can be attributed to three layers of graphemic identity: i. sub-graphemic features: serifs, roundness; ii. systematic graphical relations: majuscule vs. majuscule forms, similar relations between pairs of letters as in ‘n’ → ‘m’ ≡ ‘v’ → ‘w’, etc. iii. set of graphemes: full-fledged Roman letters borrowed on the basis of their similarity with han’gŭl characters. Note that a two-dimensional feature matrix [± form, ± meaning] analysis for a semiotic sign is not applicable here due to two reasons: (i) the [± form]-distinction is insufffijicient; we have to keep apart at least the graphemic level of existing characters and the sub-graphemic level of graphical features and ‘primitives’ on which the graphemes are based; (ii) the [± meaning]-distinction is irrelevant since the meaning of graphemes– in case of alphabet letters it is its pronunciation—was not relevant for its impact on han’gŭl reform ideas. Although discussed continuously—and sometimes vividly—online writing served mainly as a visual icon for the han’gŭl movement appearing in headlines, titles, etc. The texts written on online writing outnumber those written in online writing by far. Its peripheral status is further indicated by the fact that most proposals published in a book are contained in the appendix or the last chapter. The main object of linguists working in this fijield was the creation of a pure Korean language style and the abolition of Chinese characters for writing Sino-Korean words. Part II: Reality & Legacy After liberation the revival of Korean language in public life saw proposals for online writing by Cho Pyŏng-hŭi (1946. cf. Kim Chŏng-su 1982: 22) and Ch’oe Hyŏn-bae (1947, s. above), both similar to Ch’oe Hyŏn-bae (1926) and Kim Sŏk-kon (1932). In 1953 an offfijicial proposal based on the model of the Korean Language Society (1936, s. above) was worked out and shown to President Yi Sŭng-man (= Singman Rhee / Syngman Rhee) but fijinally rejected. The same happened to a proposal under the leadership of Ch’oe Hyŏn-bae in 1956 (Kim Min-su 1984: 263). In North Korea Kim Tu-bong had became a political force and urged the adoption of online writing until he was purged in 1958. Kim Il Sung (= Kim Il-sŏng) is said to have

124

thorsten traulsen

been talking about it as late as 1980 (Pratt & Rutt 1999: 162). Although their “revolutional spirit” would more likely have pursued radical shifts in traditional matters, the more nationalistic than socialistic North Korean leadership were careful to avoid policies in contrast with those of South Korea that could hinder future re-unifijication. But it was not only in the political fijield where online writing failed to succeed. Although, as shown above, there was an underlying ideological motivation for online writing expressed in the imitation of Western letter shapes, the reasoning pro online writing has always been based on pragmatic and scientifijic arguments. In particular, the matter of the “mechanization of han’gŭl” (han’gŭl-ŭi kigyehwa) in form of printing and typewriting seemed to be the driving force, just as Kim Sŏk-kon argued in his 1932 online writing proposal that there are voices calling for a Korean typewriter (1932: 212). But the technical and commercial breakthrough in the mechanization of han’gŭl came without online writing in 1949, when Kong Pyŏng-u’s speedtypewriter hit the market.8 Coming up with a workable Korean typewriter was not only a matter of practical need, but also of national(istic) importance, even more since Lin Yudang had just invented a Chinese typewriter in 1946. “It is a pity that our nation, having the most splendid alphabet in the world, has not yet a splendid typewriter” Kong Pyŏng-u comments on his Chinese predecessor (1949: 48). Kong Pyŏng-u gave up trying to stick to the traditional strict “ideal” square frame of a compound syllable and treated initials (i.e., consonants plus the ‘zero’ initial), medials (i.e., vowels), and fijinals (i.e., single consonants and consonant clusters) separately in only “three tiers” (sebŏl). On the contrary to other four- or fijive-tier proposals to typewriting made before (and afterwards), this method made speed-typing possible. “In the twentieth century, the age of ‘speed’ [sŭp’i-dŭ], those who use machines fijirst, will do fijine.” reads a contemporary advertisement. The price for speed-typing, however, was “abandoning-the-squareframe” (t’al-nemo-t’ŭl) of syllable compounds and having syllable blocks of diffferent size mainly depending on the existence of a fijinal (this was also true, although to a lesser degree, for the four- and fijive-tier proposals):

8  The fijirst still existing typewriter for Korean was invented by Yi Wŏn-ik around 1914 by simply replacing the English typefaces with han’gŭl typefaces. Korean was written sideways, so the page had to be turned 90 degrees to read top-down. Before, it is said that Korean labours on Hawai’i had a similar idea (Sejong Taewang Kinyŏm Saŏphoe 2000: 214). Also, the Presbyterian missionary H.G. Underwood is said to have done the same in early 1900 (id.:187).

han’gŭl reform movement in the twentieth century

125

Fig. 24. A line written with Kong Pyŏng-u’s three-tier typewriter (from Kim Chŏng-su 1982).

This font style is not pleasing to the traditional eye, but Kong Pyŏng-u argues that Roman letter shapes have undergone the same process and changed from letter shapes of the same height to shapes of diffferent height due to “scientifijic reasons” (1949: 52). “Only if we do the same scientifijic research on han’gŭl letter shapes, we can establish practical and truly aesthetic letter shapes. There is no change in the scientifijic foundations, but aesthetics change easily according to fashion and habit.” he goes on (id.). When Kong Pyŏng-u spoke of aesthetics changing easily he was right, at least as far his “abandoning-the-square-frame”-style was concerned. Born out of technical necessity, it became a font style family attracting more and more attention in the seventies (Sejong Taewang Kinyŏm Saŏphoe 2000: 261). Beginning in the early nineties graphic design made extensive use of this font style (id.): the individual letters became more spread out up to the point where it in fact becomes online writing on three tiers, visually combining both strains of han’gŭl modernization discussed so far (fijig. 25).

Fig. 25. “Abandoning-the-square-frame” (t’al-nemo-t’ŭl) font styles: from technical necessity in Kong Pyŏng-u’s type-writer (1949) to typography in the seventies, to graphic design more and more popular since the early nineties (Sejong Taewang Kinyŏm Saŏphoe 2000: 261–262). In traditional, ‘squarish’ writing this would look like:

126

thorsten traulsen

The most well-known designer for this kind of typography is An Sang-su (*1952) who created his famous An Sang-su font (An Sang-su ch’e) in 1985 (id.:180):

Fig. 26. An Sang-su font (1985, left) used for two words, the fijirst, without syllable fijinals, leaves space where the second word places its fijinals. In traditional ‘squarish’ font style this appears as follows:

What online writing and “abandoning-the-square-frame” font styles have in common is not only their roots in an intended modernization of han’gŭl, but also the linking of these activities to a leftist-nationalistic ideology. While the former served as a visual icon for the han’gŭl movement and their activities, the latter became associated with Korean reform-oriented leftist-nationalism in the latter half of the twentieth century. (However, the connection between this font-style and political protest culture in the 1980s seems to be based on type-written flyers rather than on an allusion to Colonial Period activists.) By comparing the proprietary font styles of two representing Korean quotidian newspapers this impression can be made more objective. The conservative right-winged Chosŏn Ilbo (Sino-Korean meaning “Korean Daily”) uses a “squarish” font of the myŏngjo-family, while the leftist Han Kyŏrye (native Korean meaning “Korean/Great descent”) uses a slight “abandoning-the-square-frame” font that leaves some space beneath syllable compounds without a fijinal (this efffect is strengthened in graphic design, e.g., the logo of the Han Kyŏrye):

han’gŭl reform movement in the twentieth century

127

Fig. 27. The proprietary reading fonts (not the graphic design) of the conservative Chosŏn Ilbo (top) and the leftist Han Kyŏrye newspapers (bottom). The latter leaving slightly more space beneath syllables with vertical vowel without fijinal alluding to the “abandon-the-square-frame” font style. Both fonts can be downloaded for free.

The relation between ween the Han Kyŏrye and the han’gŭl movement is going much beyond politically motivated font style and is bringing us back to the very start of the han’gŭl nationalism, that is to “The Independent” (Tongnip sinmun) from 1896 and one of its chief editors Chu Si-gyŏng. The Han Kyŏrye was founded in 1988 in the heydays of South Korean students protest movement for democratization. Contrary to the Chosŏn Ilbo, the Han Kyŏrye discards Chinese characters and uses han’gŭl only—just as “The Independent” did. In fact, the foundation of the Han Kyŏrye was inspired by “The Independent” (Vladimir Tikhonov, personal communication). Summary The impact of the Roman alphabet associated with technological development at the end of the 19th century led to a re-appraisal of phonological writing in Korea. The once prestigious Chinese logograms making the written language of the peninsula for more than 1,500 years, were now considered to be backward, at least among Korean language reformers. On the other hand, great pride was taken in the “most splendid” alphabet in the world, that is the Korean han’gŭl alphabet. The original sophisticated grammatogeny of han’gŭl (1443/4) conceived by King Sejong and his scholars strengthened this point beyond doubt and gave Korean nationalist the possibility to rank high(est) in a new world order with new cultural hierarchies. Nevertheless, linguistic reform ideas did not stop before han’gŭl. It was the syllabic compound writing that was criticized as being too cumbersome for learning, reading, writing, and printing. Modern Korean writing should be online writing (hoengsŏ / karo-ssŭgi / p’urŏ-ssŭgi). However, online

128

thorsten traulsen

writing proposals soon went beyond the level of solving the technical questions raised by Roman stimulus difffusion. Instead, they more and more tried to imitate Roman letters and their features like cursive style, serifs, minuscule vs. majuscule, up to the point where Roman and Cyrillic letter shapes were adopted to replace similar looking han’gŭl letters (‘receptive imitation’), or—as in the most “sophisticated” proposal—the borrowing was triggered by fijinding similar graphical relations in Roman letters as in han’gŭl (‘transmitting imitation’). In a schizophrenic process the pride in han’gŭl led to proposals that made han’gŭl look like a mixed Roman-Cyrillic writing.9 However, the online writing movement failed both politically as well as technologically. The fijirst technologically as well as commercially successful han’gŭl typewriter conceived by Kong Pyŏng-u (1949) proved that online writing was not necessary for typing. This was achieved by giving up the traditional “ideal” square layout of han’gŭl compound syllables and tolerate font styles that “abandon-the-square-frame” (t’al-nemo-t’ŭl). This type face family became more and more popular in typography and graphic design, i.e., outside the original context of typewriting. In its most extreme graphical usages it is indeed online writing, however on three tiers for initial consonant and vertical vowel, horizontal vowel, and fijinal. The common feature of weakening the traditional system of syllable compound writing in both online writing and “abandon-the-squareframe” font styles is accompanied by a shared political attitude that both innovations visually represent, namely a reform-oriented leftistnationalistic one. Online writing became less discussed since the nineties, probably both for technical (computer technology made the typewriting problem obsolete) and political reasons (democracy was consolidated). Neither crossword puzzles (they are in syllabic writing) nor Scrabble (assumingly absent) could help online writing sneak into everyday life. Writing across the border initiated the movement, but crossing the border in the other

9  This “serving-the-Great” (sadae) attitude can be seen in at least two other cases in the Korean history of writing: (i) the reluctance to use han’gŭl instead of written Chinese by clergy and in literary contexts, even for the transcription of foreign, e.g., Mongolian names. Here preference was given to Chinese characters used as syllabograms or even Korean-made ligatures of two of those characters to keep up a Chinese appearance of the whole text; (ii) the imitation of Manchu writing in handwritten han’gŭl in the Late Chosŏn period (1598–1910), roughly corresponding to the Chinese “Manchurian” Ch’ing dynasty (1644–1910).

han’gŭl reform movement in the twentieth century

129

direction did not take up this stimulus again: the successful export of han’gŭl to the Cia-Cia tribe in Sulawesi in August 2009 does not make use of the more simple online writing and sticks to syllabic writing. References An, Cha-san 1928. Pŏmŏ-wa Chosŏnŏ-wa-ŭi kwan’gye [The relationship between Sankrit and Korean]. Pulgyo 52: 53–58, 53: 40–46, 54: 46–48. (Reprinted in Yŏktae Han’guk munbŏp taegye [Major Works on Korean Grammar], section 3(11), pp. 302–315.) Ch’oe, Hyŏn-bae 1947 [republished 1983]. Kŭlcha-ŭi hyŏngmyŏng [The revolution of letters]. Sŏul: Chŏngŭm Munhwasa. —— 1963. Han’gŭl karo-kŭlssi tokpon [Korean online writing reader]. Sŏul: Chŏngŭmsa. Chosŏn Chŏngmunhoe (ed.) 1925. Kŭl-e tae-han mundap [Questions and answers concerning writing]. Yŏktae Han’guk munbŏp taegye [Major Works on Korean Grammar], section 3(12), pp. 3–24. Chosŏnŏ Hakhoe [Pyŏnch’ansil] (ed.) 1938. Kaejŏng-toen man’guk ŭmsŏng kiho-rŭl sogaeham [An introduction to the revised IPA]. Han’gŭl 6 (10): 25–28. Chu, Si-gyŏng 1914. Mal-ŭi sori [Speech sounds]. Chu Si-gyŏng yugo [Postumous works of Chu Si-gyŏng]. Kyŏngsŏng [i.e., Sŏul]: Chungyŏng Insŏgwan. Chung, Kei Won 1938. The origins of the Korean alphabet. Ph.D. thesis, Princeton University. Daniels, Peter 1992. The syllabic origin of writing and the segmental origin of the alphabet. In Downing, Lima, and Noonan (eds.), Linguistics of Literacy, pp. 83–110. Amsterdam: John Benjamins. Kim Chŏng-su 1982. Han’gŭl-ŭi kaebal-ŭl wi-han yŏn’gu—han’gŭl p’urŏ-ssŭgi-wa han’gŭl ŭmsŏng kiho [A study on the development of han’gŭl—han’gŭl online writing and han’gŭl sound transcription]. Han’gŭl 177: 133–177. Kim-Renaud, Young-Key (ed.) 1997. The Korean alphabet—its history and structure. Honolulu: University of Hawai’i Press. Kim, Chin W. 1997. The structure of phonological units in han’gŭl. In Kim-Renaud, YoungKey (ed.), The Korean alphabet—its history and structure, pp. 145–160. Honolulu: University of Hawai’i Press. Kim, Min-ju (ed.) 1992. Chu Si-gyŏng chŏnsŏ [Complete works of Chu Si-gyŏng]. Sŏul: T’ap. Kim, Min-su 1984. Kugŏ chŏngch’aengnon [Korean language policies], pp. 254–268. Sŏul: T’ap. Kim, Sŏk-kon 1932. Han’gŭl karo-ssŭgi [Han’gŭl online writing]. Tonga Ilbo [East Asia Daily] June 14 to July 20, 1932. (Reprinted in Yŏktae Han’guk munbŏp taegye [compilation of sources], section. 3, vol. 10, pp. 209–47.) Kim, Tu-bong 1922. Kiptŏ Chosŏn malbon [Deepened and amended Korean grammar]. Shanghai: Saegŭlchip. Kim, Yun-gyŏng 1937. Han’gŭl karo-ssŭgi-ŭi sajŏk kwanch’al [A historical survey of han’gŭl online writing]. Han’gŭl 5(2): 8–15. King, Ross 1997. Experimentation with han’gŭl in Russia and the USSR, 1914–1937. In KimRenaud, Young-Key (ed.), The Korean alphabet—its history and structure, pp. 219–261. Honolulu: University of Hawai’i Press,. —— 1998. Nationalism and language reform in Korea—the questione della lingua in precolonial Korea. In Pai, Hyung Il & Tangherlini, Timothy R. (eds.), Nationalism and the construction of Korean identity, pp. 33–72. Berkeley: University of California. Kong, Pyŏng-u 1949. Nae-ga koan-han ssangch’otchŏm han’gŭl t’ajagi [The fijirst han’gŭl typewriter that I proposed]. Han’gŭl 14(1): 48–54.

130

thorsten traulsen

Pak, Hwan 2008. Rŏsia Hanin yujŏk tapsagi [An essay on historical remains of Koreans in Russia]. Sŏul: Kukhak charyowŏn. Pratt, Keith & Rutt, Richard 1999. Korea—a historical and cultural dictionary. Richmond: Curzon. Ri, P’il-su 1923. Chŏngŭm munjŏn [Grammar of correct sounds]. In Yŏktae Han’guk munbŏp taegye [Major Works on Korean Grammar], section 1, vol. 13, pp. 1–35. Sejong Taewang Kinyŏm Saŏphoe [Han’guk Kŭlkkol Kaebarwŏn] 2000. Han’gŭl kŭlkkol yongŏ sajŏn [Han’gŭl letter shape terminology dictionary]. Sŏul: Sejong Taewang Kinyŏm Saŏphoe. Yi, Kyu-yŏng 1911–13. Onkat kŏt [All kind of things]. In Kim, Min-ju (ed.) 1992. Chu Si-gyŏng chŏnsŏ [Complete works of Chu Si-gyŏng], vol. 4, pp. 461–506. Sŏul: T’ap.

THE CHARACTER OF THE INDIAN KHAROṢṬHĪ SCRIPT AND THE “SANSKRIT REVOLUTION”: A WRITING SYSTEM BETWEEN IDENTITY AND ASSIMILATION Ingo Strauch Introduction The history of writing in India is extraordinary from diffferent points of view. First, there is the fact that writing as a cultural technique was introduced into Indian culture at a rather late point in its history. It was long after large amounts of its literature had already been created. Not only the Veda but also the late Vedic literature and even considerable portions of the Buddhist and Jain canons were already in existence when India saw its fijirst attempts to fijix language to a writing system.1 Remarkably, this introduction did also postdate the early Indian state formation processes in the 5th century BC, and consequently it was also later than the urbanization which accompanied this state formation in the Ganges valley (cf. Strauch 2005). The focus of my paper will be dedicated to another peculiar feature of the Indian history of writing. From its very beginning two rather different scripts were in use: the Kharoṣṭhī, a script which prevailed in the Indian North-West, i.e. modern North-West Pakistan, and spread from there to Northern Bactria and Central Asia, and the Brāhmī which began

1  The question of the introduction of writing in India has been controversially discussed. An excellent survey of this academic debate and a plausible scenario of this process was presented by Falk (1993). Other valuable contributions concerning this problem are von Hinüber (1990) and Falk (1996). Although it is probable that the introduction of Kharoṣṭhī preceded that of Brāhmī for some decades there is no positive evidence for the use of either of these scripts before the time of Aśoka. Any suggestions about a gradual development of Kharoṣṭhī in the centuries before Aśoka and a reconstruction of this process (e.g. Glass 2000: 11–20) must remain highly speculative. For the suggested very small time gap between the emergence of Kharoṣṭhī and Aśoka see Falk (1993: 103–105). According to Salomon, “there is no clear evidence to allow us to specify the date of the origin of Kharoṣṭhī with any more precision than sometime in the fourth, or possibly the fijifth, century B.C.” (1998a: 46). Recent discoveries of presumably earlier Brāhmī texts from Śrī Laṅkā still wait for further confijirmation of their suggested dating between the 6th and 4th centuries BC (see Coningham et al. 1996). Since this article will concentrate on the fijinal phase of Kharoṣṭhī, the discussion on its introduction is of minor relevance.

132

ingo strauch

its career in the East of the Indian subcontinent and conquered almost all of India until it replaced even the Kharoṣṭhī in its mother-land after the 3rd c. AD. It is the Brāhmī script which became the “mother” of all modern South-Asian scripts and of many scripts in South-East Asia. The Kharoṣṭhī remained a footnote and left no further traces in the writing systems of the region. Nonetheless, it was a rather influential script which not only left hundreds of epigraphs but also a considerable corpus of Buddhist manuscripts which belong to the earliest witnesses of Buddhist literature in general.2 In recent years our knowledge about this script and its use increased considerably due to the discovery of a series of new Kharoṣṭhī manuscripts which are presently being studied in Seattle and Berlin.3 Why this script died out, is still a matter of academic debate. Only recently, Richard Salomon, one of the leading experts in Kharoṣṭhī studies, expressed the view that “the decline of the Kharoṣṭhī in its homeland was closely connected with, if not directly caused by, the collapse of the Kuṣāṇa dynasty.” He attributes the Kharoṣṭḥī case to those “cases of script disappearance that are directly attributable to dynastic changes or declines (Salomon 2008b: 149).” According to Salomon, the Kharoṣṭhī “could in theory have fulfijilled the role of a Pan-Indian script, that, as the accidents of history had it, actually fell to Brāhmī and its derivatives (2008b: 144).” It is the main aim of this paper to demonstrate a diffferent approach to this phenomenon in the Indian history of writing and to show that the decline of Kharoṣṭhī can alternatively be interpreted as the direct result of a cultural and linguistic shift in the communities where it was used, and not—or at least not mainly—as the consequence of a political event. Taking up the general title of this book this shift could be characterized as a forced and intentional, but eventually unsuccessful border-crossing by which the Kharoṣṭhī script was supposed to adjust itself to another language, namely Sanskrit. The paper will be divided into two parts. The fijirst section describes the character of the script, particularly with regard to its suitability to write Sanskrit. Since the missing success of the Kharoṣṭhī is also the result of a

2  An easy survey about all material available in this script can be found in the “Catalog of Kharoṣṭhī Inscriptions” (CKI) and the “Catalog of Kharoṣṭhī Manuscripts” (CKM) which are both available on the homepage of the “Dictionary of Gāndhārī”, a project by Stefan Baums and Andrew Glass (www. Gandhari.org). The page also provides an extensive bibliography. 3  See for more information about the manuscripts studied in both projects Salomon 1999, Allon 2007, Strauch 2008.

indian kharoṣṭhī script and the “sanskrit revolution”

133

competitive situation between both Indian scripts and their cultural contexts this section will also shortly refer to the characteristics of Brāhmī. The second part will illustrate diffferent attempts of Kharoṣṭhī to adjust itself to the introduction of Sanskrit as a new literary language and lingua franca in the Gandhāra area and its relationship to the Brāhmī script which accompanied the advent of Sanskrit. It should be stressed that this paper does not aim at a comprehensive study on the subject. It tries to introduce a new perspective on the problem in a more essayistic form, including some new data which are based on the author’s recent research on a collection of Kharoṣṭhī birch-bark manuscripts (Strauch 2008). Since the fijield of Kharoṣṭhī studies has recently been much in transition due to the ongoing discovery of new texts and inscriptions, the general and comprehensive history of Kharoṣṭhī remains to be written. The Character of the Kharoṣṭhī Script The beginnings: Brāhmī and Kharoṣṭhī under Aśoka The fijirst safely datable monuments of the Kharoṣṭhī are associated with the reign of the Indian ruler Aśoka (3rd c. BC). On behalf of his imperial order, rocks and pillars all over the South-Asian subcontinent were inscribed with what one might call a kind of state ideology which is partially based on Buddhist ethical conceptions.4 While for the majority of texts the probably newly developed Brāhmī script was used, those inscriptions which are located in the extreme North-West are written in Kharoṣṭhī. Contemporary to the Aśokan Kharoṣṭhī epigraphs are some texts which continue the use of the earlier Greek and Aramaic alphabets and languages in the Indian North West (Falk 2006: 241–253). These contemporary writing cultures also indicate the most probable candidates for the stimulus to create a script which is capable of fijixing an Indian language. According to the prevailing theories, the Kharoṣṭhī script was developed on the base of the Aramaic alphabet as used during the Mauryan period in North-Western India.5 This presumable prototype is not only responsible for the shape of

4  A comprehensive survey of the Aśokan epigraphs and a bibliography are now available with Falk 2006. 5  The complex relationship between Aśokan Kharoṣṭhī and Aramaic has been dealt with by Falk (1996). The Semitist’s point of view was expressed by Voigt (2005), whose remarks provide valuable additional data but sufffer from the author’s lacking familiarity with the Kharoṣṭhī script and Gāndhārī language.

134

ingo strauch

many individual letters but also for the writing direction of the Kharoṣṭhī from right to left. Despite these parallels, the Kharoṣṭhī has to be defijined as a newly created script with a distinctively diffferent character. This fijirst phase of Indian writing is not only distinguished by the multitude of scripts but also by the fact that its texts were composed in different dialects. Consequently, beside their doubtless ritual function to mark the extant of the ruler’s actual or asserted influence, these edicts were clearly meant to be understood by the local communities. For this purpose Aśoka ordered them to be translated into various Middle Indian dialects. Among these translations only those in the North-Western language were written down in Kharoṣṭhī. This speaks not only in favour of the assumption that this script was already in use at the time of Aśoka, but also that it had been intentionally designed for the local language of the region, the so-called Gāndhārī. The Brāhmī, however, fulfijilled from its very beginning a much more universal function and was designed to express diffferent local dialects of Northern India. What both scripts shared, however, is the fact that none of them was created to write down a Sanskrit text. This had a direct impact on their inventory of signs. Kharoṣṭhī and the Sanskrit varṇamālā Usually, Indian alphabets are arranged according to the so-called varṇamālā sequence: an inventory of sounds which was created on the basis of Sanskrit phonology as early as in the 4th c. BC, i.e. before the introduction of writing. It classifijies the sounds of the language according to their physiological features into diffferent groups (cf. table 1). A script which could claim to match these Sanskritic phonemes would be expected to provide graphemes for all these sounds. Since not all of them are needed for Middle Indian, the earliest Brāhmī and Kharoṣṭhī have only a restricted inventory of signs. The table below contrasts the traditional varṇamālā alphabet and those sounds which were not represented in the scripts in their initial stages. It should be stressed that in the case of Kharoṣṭhī this survey mainly relies on the conventional phonetic evaluation of the respective graphemes, which is in many cases, however, far from certain.6

6  For a promising new approach to the phonology of Gāndhārī including a discussion of the phonetic value of Kharoṣṭhī signs see now Stefan Baums (2009: 110–200). The later use of the Kharoṣṭhī graphemes for writing Sanskrit, however, shows that the Kharoṣṭhī signs could indeed carry the phonetic value implied by the varṇamālā (cf. below, table 3).

indian kharoṣṭhī script and the “sanskrit revolution”

135

Table 1. The Sanskrit varṇamālā alphabet and its relation to early Brāhmī and Kharoṣṭhī  not in Kharoṣṭhī

 not in Brāhmī and Kharoṣṭhī Vowels Simple vowels short

a

i

u

r̥

long

ā

ī

ū

r̥ ̄

short

e

o

long

ai

au

ḷ

Diphthongs

Consonants Explosives

voiceless

voiced

Nasals

non-aspirated

aspirated

non-aspirated

aspirated

Velars

ka

kha

ga

gha

ṅa

Palatals

ca

cha

ja

jha

ña

Retroflexes Dentals Labials

ṭa ta pa

ṭha tha pha

ḍa da ba

ḍha dha bha

ṇa na ma

Semisonants

ya

ra

la

va

Sibilants

śa

ṣa

sa

Spirant

h

Vowel modifijicators

ṃ Anusvāra

Consonant modifijicators

Virāma

ḥ Visarga

136

ingo strauch

Relying on this varṇamāla based comparison one can easily get the impression that both alphabets contain by and large the same inventory of signs. Their coincidences can be explained on the basis of the common Middle Indian phonology: both lack the velar ṅa and the special aspiration sign Visarga, and both lack the signs for sonantic r̥ /l ̥ and the diphthongs ai and au. Beside this seemingly common inventory both alphabets also share their abugida character. Every consonantal sign of the script includes a subsequent vowel (cf. Daniels & Bright 1996: 4), in the case of the Indian scripts a short a. To change the quality or quantity of this basic vowel one has to add another diacritical sign. The inherent vowel expressed by a sign makes it also necessary to signify vowelless consonants. This had to be done either by the graphical instrument of sign combinations, the so-called ligatures, or—if the consonant appeared at the end of a word— by a further diacritical sign. Since the Middle-Indian languages are characterized by a strong tendency to assimilate diffferent and to omit fijinal consonants, both early variants of Kharoṣṭhī and Brāhmī contained a very limited choice of ligatures and no special diacritic mark for a fijinal consonant. In addition, both scripts generally refrain from designating geminated consonants, in which the Middle Indian languages are extraordinarily rich due to the described tendency of consonant assimilation.7 Beside these shortcomings which had to be overcome to make both alphabets fijit for Sanskrit, the Kharoṣṭhī shows a further signifijicant diffference: It is unable to signify the vowel quantity of ā̆, ī ̆ and ū̆. It is rather improbable that the Gāndhārī speakers did not know about this distinction. Thus this fact can be best explained as a structural inconsistency which was probably inherited from the Semitic model of the script. It is mainly this diffference, which marks the Brāhmī as the more developed of both scripts, which is better adjusted not only to the requirements of Middle Indian phonology but also to those of Sanskrit.

7  A special position among the consonant clusters is held by the combination of consonants with the semivowel r. Since the local dialect of the Indian North-West obviously did not assimilate this sound in post- and preconsonantal position, the Kharoṣṭḥi knew special diacritics for these combinations from the very beginning. Although this sound combination was obviously also known in other Middle Indian dialects, its adequate realization in the Aśokan Brāhmī caused a few problems which were, however, quickly overcome in the subsequent century (cf. Dani 1963: 54f.).

indian kharoṣṭhī script and the “sanskrit revolution”

137

Kharoṣṭhī and the Arapacana As useful as the varṇamālā arrangement might be for a direct comparison of both scripts, it hides some of the characteristic features of Kharoṣṭhī and gives the impression that with some exceptions both alphabets share a common character. This impression quickly disintegrates when we consider the sign inventory of the Kharoṣṭhī on the base of its own alphabet, which is known today by the name Arapacana because it starts with the fijive letters a-ra-pa-ca-na. In recent East-Asian Buddhism this alphabet is associated with the Buddha Mañjuśrī, and its name—Arapacana—is regarded as a kind of mantra, i.e. a magical spelling (Salomon 1990: 255). In a much earlier period the alphabet was also used as a kind of memorizing device: important categories of Buddhist doctrine were memorized in the sequence of the alphabet. Remains of this use can be observed in quite a large number of Buddhist texts, where various lists of Buddhist terms contain passages which can be traced back to the sequence of the Arapacana (Pagel 2007: 18–38). We do not know how old this alphabet is and since when it had been used for the arrangement of the Kharoṣṭhī sign system. The fact that it contains some signs which are obsolete in the mature phase of Kharoṣṭhī (no. 28, no. 40) seems, however, to speak in favour of a quite early date of the Arapacana. Until most recent time the inventory of the alphabet had to be reconstructed on the base of few Sanskrit texts, like the Śatasāhasrikā and Pañcaviṃśatikā Prājñāpāramitā or the Lalitavistara, which contain an almost complete but sanskritized version of this alphabet in Brāhmī writing (Brough 1977, Salomon 1990). In many cases it was not possible to determine which concrete Kharoṣṭhī sign hid behind the respective Sanskrit-Brāhmī “transcription”. Only recently, several incomplete variants of the Arapacana alphabet written in Kharoṣṭhī could be identifijied. The oldest of them was discovered on an ostracon in the course of excavations of a Buddhist monastic complex at Kara-Tepe, near Termez, Uzbekistan (Salomon 2004, Fussman 2011: 87, 107 KT). According to its archaeological context its date can be fijixed into “the late fijirst or early second centuries A.D.” (Salomon 2004: 45). Another part of the Arapacana alphabet was found among the inscribed wooden tablets from Niya in Chinese Turkestan, which can be dated into the 3rd, early 4th centuries AD (Salomon 1990). This rather limited evidence is supplemented by the widespread use of Arapacana letters as location markers in Gandhāran art and architecture. Many objects—reliefs, sculptures, etc.—bear small

138

ingo strauch

Kharoṣṭhī signs which indicate their arrangement in a certain complex (Salomon 2006, Koizumi 2007). All this evidence allowed a rather reliable reconstruction of the alphabet up to its 35th letter (cf. table 2). Fortunately, among a collection of manuscripts which was discovered in the Bajaur district, not far from the Afghanistan-Pakistan border, and is currently being studied at Freie Universität Berlin, a text could be identifijied which contains a compilation of verses which are arranged in the order of this alphabet (= BajC 5, cf. Strauch 2008: 121–123). Each half of every verse begins with the same letter. On the basis of this text and the incomplete versions which were previously known it is now possible to establish the complete inventory of signs as listed in the Arapacana alphabet. The following table (table 2) will demonstrate the inventory of signs as given in the Bajaur fragment 9 and its relation to other versions of the Arapacana alphabet. Arapacana signs which are not part of the varṇamālā alphabet are left unshaded in columns 1 and 6. (Section 1). Moreover, the table will show the varieties of secondary modifijications attested in various sources of Kharoṣṭhī writing (Section 2). A third section will illustrate the use of the basic and modifijied signs in a Kharoṣṭhī text from Bajaur which is composed in Sanskrit (= BajC 9). The Arapacana alphabets in the table are reconstructed according to the following publications: – Brāhmī representatives from Skt. texts: Salomon 2004: 47 – Niya tablet: Salomon 1990: 265–268, fijig. 7 (with the exception of no. 21 (śpa), no. 28 (ṭ́a) and no. 35 (ṭ́ha) which were corrected on the basis of the digital image of the plate (available at http://idp.bl.uk/database/large.a4d ?recnum=16781&imageRecnum=20818) for spa, (tha), and [ṭha] – Kara-Tepe ostracon: Salomon 2004 – Location markers: 1–17 (a-ma): Zar Dheri (Koizumi 2007), except 13 ( ya) (Salomon 2006: 199), 18 ( ga): Nimogram (Salomon 2006: 200), 24 (kha): Butkara (ibid.), 26 (sta): “Chorasan, Gandhara” (ibid.). The Bajaur signs are extracted from the verse initials of Bajaur fragment 5, except when preceded by *. In these cases the signs had to be taken from other portions of the text. Most of the modifijied signs are extracted from Glass 2000. Their exact provenance is indicated in the following annotated survey. This table does not aim at a complete inventory of modifijied signs and is meant only for illustration purposes. The actual—recently attested—number of modifijications is higher than shown here. A more complete survey has to be undertaken after most of the recently studied Kharoṣṭhī manuscripts are edited. It is possible that some of the signs identifijied by Glass as phonetically relevant graphemes are mere ornamental varieties of the basic sign. If the character of a modifijication is unclear, this is indicated by *.

indian kharoṣṭhī script and the “sanskrit revolution”

139

– *p̱a: Glass 2000: 84 (Gardner plate 30, no.3). Although Glass mentions the variant p̱a (transliterated as ṕa) among the letters composed with the help of the respective syllable modifijicator (“Cauda”) (2000: 136), his chapter on the variants of the sign pa contains only one letter of this type which is considered by him as a footmark (2000: 84). It is generally diffijicult to distinguish the phonetically relevant substroke from the merely graphical footmark of almost identical shape. – c̄ a: Glass 2000: 62 (Niya documents) – n̄a: Glass 2000: 82 (Khotan Dharmapada) – ḏa: Glass 2000: 79 (British Library scribe 21) – ḍ̱a: Glass 2000: 72 (Niya documents). Although Glass is referring to this variety in his chapter on modifijiying signs (2000: 136), he refuses this identifijication in his chapter on the sign ḍa and suggests considering it as one of the phonetically irrelevant footmark variants (2000: 73). For sake of completeness we will, however, pertain to distinguish this variant from its basic sign. Rapson et al. (1920–29) transliterate this letter as ḍ́ a. – ṣ̄a: Glass 2000: 103 (British Library scribe 9) – ṣ̱a: Bajaur Collection scribe 4 – va: Glass 2000: 98 (Niya documents). As Glass points out, this letter is already found on the coins of Vima Kadphises, where his name is spelt vima. – ṯi: Glass 2000: 77 (Bajaur casket) – *y̱i: Glass 2000: 94 (Bajaur casket). It is not clear whether this variant really represents a modifijied consonant. Again Glass mentions this letter among the examples for the consonant modifijier (2000: 136), he considers it in another chapter as a footmark variety of the letter ya (2000: 94). – ḵa: Glass 2007: 95 (Senior scribe) – k:̄ Bajaur Collection scribe 3 – s̄e: Glass 2000: 109 (British Library scribe 14) – s̱ a: Glass 2000: 107 (British Library scribe 9). It is not clear whether the variants (Wardak vase) and of some British Library scribes represent the same sign. They are regarded by Glass as early variants of the letter s̱ a. In one of the BajC manuscripts, however, both variants were used side by side, indicating perhaps that they were perceived as diffferent graphemes. The same graphical distinction between the ordinary s̱ a and this under-bent sa is also found in the Arthapada manuscript of the “Split Collection” where both signs perhaps even designate diffferent phonemes. Cf. Falk 2011b: 14f. – m̄ a: Glass 2000: 93 (British Library scribe 8) – *mi: Glass 2000: 92 (Wardak vase). Again the status of this sign is unclear. Like the p̱a, the ḍ̱a and the y̱i it was listed by Glass among the modifijied signs (2000: 136), but considered in the description of letters as a footmark variety (2000: 92). – g̱i: Glass 2000: 59 (Khotan Dharmapada) – ḡa: Glass 2000: 58 (Khotan Dharmapada) – th̄a: Bajaur Collection scribe 9

140

ingo strauch – j̄a: Glass 2000: 65 (British Library scribe 9) – j̱a: Glass 2000: 66 (Bajaur casket) – śā : Glass 2000: 99 (British Library scribe 10). According to Glass, “the superscript line with śa is not thought to signify a diffferent phoneme, but rather to distinguish śa from ya which can be almost identical” (2000: 99). Although this interpretation is quite plausible, it seems advisable to consider this sign for the time being as a separate and phonetically relevant variant of the basic grapheme śa. – ś̱a: Glass 2000: 100 (Bajaur casket) – kṣ̄a: Glass 2000: 116 (Niya documents) – ñ̄: Bajaur Collection scribe 9 – h̄e: Glass 2000: 113 (Niya documents)

a

ra

pa

ca

na

la

da

ba

ḍa

ṣa

va

ta

1

3

4

5

6

7

8

9

10

11

12

Sanskrit-Brāhmī Niya tablet representatives

2

No.

1 Arapacana alphabet- Basic signs

ta

[va]

[ḍa]

ba

da

la

[na]

[ca]

[pa]

Kara Tepe ostrakon

ta

va

ṣa

ḍa

ba

da

la

na

ca

pa

ra

a

Location markers

ta

va

ṣa

ḍa

ba

daṃ

la

na

ca

pa

ra

*

Bajaur fragment BajC 5

ṣ̄a

ṯi

va

ṣa̱

ḍ̱a

ḏa

ta

va

ṣa

ḍa

ba

da

la

na/ṇa

n̄ a

pa

ra

a

Basic signs

ca

*p̱a

Extension below

śca

Modifijied signs

3 Sanskrit reflexes (BajC 9)

c̄a

Bar above

2 Modifijications

Table 2. The basic and modifijied signs of the Kharoṣṭhī Arapacana alphabet and their Sanskrit reflexes in BajC 5

indian kharoṣṭhī script and the “sanskrit revolution” 141

ya

ṣṭa

ka

sa

ma

ga

tha

ja

śva

dha

śa

kha

13

14

15

16

17

18

19

20

21

22

23

24

kha

śa

dha

śpa

(ja)

tha

( ga)

(ma)

[sa]

Sanskrit-Brāhmī Niya tablet representatives

No.

1 Arapacana alphabet- Basic signs

Table 2 (cont.)

kha

śa

dha

[ś..]

ja

tha

ga

ma

sa

ka

[ṭha]

ya

Kara Tepe ostrakon

kha

ga

ma

sa

ka

ṭha

ya

Location markers

kha

śa

dha

śpa

ja

tha

ga

ma

sa

ka

ṭha

ya

Bajaur fragment BajC 5

ś̄ a

j̄a

ś̱a

ja̱

g̱a

ḡi th̄a

*mi

s̱a

ḵa

*y̱i

Extension below

m̄ a

s̄e

kā

Bar above

2 Modifijications

kha

śa

dha

—

tha/ stha ja

ga

ma

sa

ṣṭa/ ṣṭha ka

ya

Basic signs

stha

Modifijied signs

3 Sanskrit reflexes (BajC 9)

142 ingo strauch

kṣa

sta

jña

rtha

bha

cha

sma

hva

tsa

gha

ṭha

ṇa

pha

25

26

27

28

29

30

31

32

33

34

35

36

37

[ṭ́ha]

[gha]

tsa

vha

spa

cha

bha

ṭ́a

ña

sta

kṣa

Sanskrit-Brāhmī Niya tablet representatives

No.

1 Arapacana alphabet- Basic signs

Table 2 (cont.)

cha

bha

ṭ́a

[ña]

[sta]

kṣa

Kara Tepe ostrakon

sta

Location markers

pha

ṇa

ṭ́ha

gha

tsa

vha

spa

cha

bha

ṭ́a

ña

sta

kṣa

Bajaur fragment BajC 5

ñ̄

kṣ̄a

Bar above

2 Modifijications Extension below

pha

ṇa/na

—

gha

tsa

—

—

cha

bha

—

jña

sta

kṣa

Basic signs

jña

kṣa

Modifijied signs

3 Sanskrit reflexes (BajC 9)

indian kharoṣṭhī script and the “sanskrit revolution” 143

Sanskrit-Brāhmī Niya tablet representatives

ska

ysa

śca

ṭa

ḍha

ha

No.

38

39

40

41

42

*

1 Arapacana alphabet- Basic signs

Table 2 (cont.)

Kara Tepe ostrakon

Location markers

*ha

ḍha

ṭa

ća

za

ḱa

Bajaur fragment BajC 5

h̄e

Bar above

2 Modifijications Extension below

ha

—

—

—

—

ska

Basic signs

Modifijied signs

3 Sanskrit reflexes (BajC 9)

144 ingo strauch

indian kharoṣṭhī script and the “sanskrit revolution”

145

It might be worth noticing that the versions of this alphabet represented here—although they originate from quite diffferent regions and cultural contexts of the Kharoṣṭhī using area—are remarkably consistent. This clearly speaks in favour of a coherent use and commonly accepted shape of the Arapacana alphabet throughout the attested history of its use. If we compare this repertoire with the varṇamālā, we observe: 1. Nearly all phonemes of the reduced Middle Indian variant of the varṇamālā as represented in the Aśokan scripts are part of the Arapacana alphabet. They are indicated here by shaded cells. The exceptions are the phonemes jha and ha which are not part of the alphabet. In the case of the aspirated jha it is quite possible that it was unknown to the Gāndhārī language.8 The reason for the absence of ha is however a mystery. It is clearly present in the earliest varieties of Kharoṣṭhī. A not very convincing explanation could point to the fact that some dialectical variants of Gāndhārī seem to reduce the Old-Indian phoneme h to y or hiatus (cf. Brough 1962: 92f.). Accordingly, the Arapacana could have been developed in a region which did not know the voiced glottal sound h. 2. In comparison to the varṇamālā arrangement the Arapacana alphabet contains a considerable number of additional letters (here left unshaded). These additional signs can be divided into diffferent categories: A—Some of these signs can be explained as representations of sounds which are only found in the Gāndhārī speaking area for which the Kharoṣṭhī was designed. To this category belong the signs for vha (no. 32), which most probably determines a voiced or voiceless spirant like v/f, and za (no. 39), a voiced sibilant which is absent from other Indian dialects. B—The phonetic value of the remaining signs is less clear. Although the graphical shape of some of them seems to suggest that they represent either sound combinations (no. 31 spa, no. 33 tsa) or modifijications of basic phonemes (no. 26 sta, no. 28 ṭ́a, no. 35 ṭ́ha, no. 38 ḱa, no. 40 ća), the respective transliterations are mere reconstructions and do not reproduce their actual phonetic value.

8  In earlier transliteration systems the Kharoṣṭhī sign za was transliterated as jha. For its now commonly accepted transliteration cf. Glass 2000: 110.

146

ingo strauch

Without entering the discussion of the complicated and not yet completely understood phonology of the Gāndhārī language and its relation to the orthography of the Kharoṣṭhī script, we may restrict ourselves here to the observation that the Arapacana alphabet contains eleven letters which are not part of the Sanskrit phonological system as displayed in the varṇamālā and are thus strictly taken not necessary to write Sanskrit. At the same time these “superfluous” letters form an indispensable part of the Kharoṣṭhī alphabet. Regardless of their original phonetic value all of the signs of the Arapacana alphabet are perceived as basic letters which can be subject to further modifijications, like the designation of the vowel value and its nasalization by a set of diacritics. Another important category are special diacritics for the pre- and postconsonantal semivowels -y-, -r-, -v-. Obviously, consonant clusters with these semivowels occurred in at least some varieties of Gāndhārī and are therefore part of the Kharoṣṭhī script from its very beginning (cf. table 3).9 In addition to this initial inventory the Kharoṣṭhī scribes developed a method to further modify the original value of the basic signs by attaching Table 3. The vowel and semivowel diacritics of conventional Kharoṣṭhī Vowel diacritics ka

ki

ku

ke

vya

sva

ko

paṃ

Semivowel diacritics rka

kra

9  All sample letters are extracted from the Bajaur manuscript BajC 2, with the exception of vya (BajC 9r).

indian kharoṣṭhī script and the “sanskrit revolution”

147

diffferent diacritics. The structural basis for this approach can already be observed in the fundamental sign inventory as displayed in the Arapacana. Here we can discern sign groups like pa: pha (no. 3: no. 37), ca : ća (4: 40), va: vha (11: 32), ḍa: ṭha : tha: sta : ṭ́a : ṭ́ha : ṭa (9: 14 : 19 : 26: 28 : 35 : 41), ka: ḱa (15 : 38) or ga : gha (18 : 34), where phonetic relationship is clearly expressed by the graphical modifijication of basic signs. It is remarkable that in cases where we can clearly distinguish the basic sign from its modifijied variant the basic form regularly precedes the modifijied one in the sequence of the alphabet. It is therefore possible that the order of the Arapacana letters reflects the historical sequence of their creation. Thus it can hardly be seen as a mere coincidence that most of the signs for aspirated sounds, which were probably originally unknown to the Gāndhārī language, are in the second half of the alphabet. I call this process which took place until the alphabet as such was closed for further additions of signs “primary modifijication”. It is difffijicult to say, when this process came to an end, i.e. when the Arapacana alphabet had reached its complete shape. It cannot be excluded that the alphabet was not yet complete by the time of Aśoka. At least some of its basic signs (no. 21: śpa, no. 32: vha, no. 33: tsa, no. 38: ḱa, no. 40 : ća) are obviously absent from the inventory of the Aśokan Kharoṣṭhī as represented in the rock edicts at Shahbazgarhi and Mansehra. However, like any argumentum ex silentio this absence cannot prove that other unattested varieties of the script did not contain these missing signs. It is therefore possible—though hypothetical—that Glass’ statement according to which “(b)y the time of the Aśokan inscriptions, the Kharoṣṭhī alphabet was complete” (2000: 20) is correct. In the centuries following Aśoka this primary modifijication was supplemented by two types of secondary modifijiers—a horizontal bar above the letter and an extension added to the foot of the letter (cf. above table 2). The phonetic value of these modifijications is not completely clear, and according to the present state of research their use was by no means consistent. Concerning the fijirst variety, however, there now seems to be good evidence that it mostly indicates the duplication—or prolongation—of a consonantal sound. It must be stressed that it shares this function with other graphical devices. Thus a preconsonantal r and a post-consonantal v was also occasionally used to designate consonant clusters, including geminated consonants (cf. Baums 2009: 198, table 45). Less clear is the function of the underlining variety which is sometimes difffijicult to distinguish from an ornamental foot-mark without phonetic meaning. In many

148

ingo strauch

cases it can be interpreted as a marker of a fricative pronunciation, but other types of modifijication are not excluded.10 The number of true combined letters which use two or more basic signs for designating consonant clusters is very limited. According to the hypothesis suggested here they are originally alien to this script and belong to the later phase of Kharoṣṭhī writing which experienced strong influence from the Brāhmī writing system (cf. below). This detailed description of the alphabet and its use was intended to show that the Kharoṣṭhī by its nature was not phonological like the Brāhmī but phonetic. Throughout its history it showed a strong tendency to depict diffferent allophones by diffferent or modifijied letters. This resulted in a rather large diversity of signs which hardly corresponded to the phonological structure of the language the script designated. This phonetic character is closely connected with and still complicated by the fact that the Kharoṣṭhī shows very weak tendencies towards a standardization of its orthography. Thus we easily fijind for one lexeme a multitude of diffferent spellings. Although it is not excluded that these various spellings reflect diffferent phonetic realizations of the same word, we have to consider that they might simply be caused by diffferent orthographical conventions, including historical spellings. This results in considerable synchronic and diachronic diversity of orthographical usages. As an example for this feature one may cite the frequently attested word derived from Old Indian pratiṣṭhāpita “erected”. Its orthographical reproductions in the Kharoṣṭhī records11 include pratistapita (CKI 155), praṯiṭhap̱iṯa (CKI 328), pratiṭhavita (CKI 257), praṯiṭhaviṯo (CKI 48), pratiṭhavida (CKI 243), pratiṭhaviḏa (CKI 32), pratiṭ́ havidu (CKI 405), praṯithaviṯe (CKI 176), pratithavito (CKI 65), pratistavitae (CKI 403), pradiṭhaviḏa (CKI 464), pradistavita (CKI 60). Although such orthographical variants might provide valuable information about the dialectical variance in the Gāndhārī speaking area, they are rather obstructive in terms of standardization which is, however,

10  For its assumed function as fricative marker cf. Glass 2000: 136f. More diffferentiated is the interpretation of this graphical device as suggested by Baums (2009). While k “indicates [j] derived from a velar” (140), ḍ̱ as used in the Niya documents and by British Library scribe 14 indicates the sound [ɽ] (141), d could designate a fricative (141) and s̱ is said to indicate “the change from [s] to [z]” (150). 11  The inscriptions are cited according to the “Catalog of Kharoṣṭhī Inscriptions” (CKI), see above fn. 2. Cf. there for detailed bibliographical references.

indian kharoṣṭhī script and the “sanskrit revolution”

149

necessary if a script is intended to be used for other languages than that it was created for. The “Sanskrit Revolution” The Aśokan inscriptions and all other epigraphical documents in the subsequent two centuries were composed in Middle Indian languages. Not a single written Sanskrit text of this period has been found so far. This fact can at least partially be explained by the adverse attitude of the Brahmins towards the newly introduced writing culture. Thus still the Mahābhārata, a work which was composed in the three centuries before and after Christ (13.24,70) states: vedavikrayiṇaś caiva vedānāṃ caiva dūṣakāḥ vedānām ̣ lekhakāś caiva te vai nirayagāminah ̣ Those who sell the Veda, who spoil the Veda, who write down the Veda will certainly go to hell.

Writing was regarded by the Brahmins an inappropriate means to preserve and transmit their religious texts. This attitude was mainly due to their monopolistic position in the sphere of religious literature. It was one of their main tasks to preserve the textual tradition and to apply it in ritual contexts. Any attempt to democratize this sphere—and script is of course an instrument in this direction—would challenge this position and was consequently rejected. It is therefore not surprising that the growing writing culture was fijirst adapted by other social and religious groups. The most important among them were the Buddhists. The majority of inscriptions which are attested between Aśoka and the 1st c. AD consequently belong to Buddhist monasteries and record the dedication of buildings or sculptures. The persons who left these epigraphs help to determine the social strata which used writing. They belonged either to the ruling aristocratic class or to the class of merchants and artisans. The language policy of the Buddhists promoted the development of Middle Indian dialects as literary languages. They translated their literature into the dialects spoken in the areas of their activities. Although in the fijirst centuries following Aśoka the texts were mainly orally transmitted the Buddhists soon started to make use of the new cultural technique of writing. The exact date of this event is in most cases unknown, but the Pāli canon e.g. is said to have been written down in the middle of the 1st century BC (von Hinüber 1990: 63–66). The newly discovered

150

ingo strauch

manuscripts from Gandhāra can be dated slightly later, into the period between the 1st c. BC and the 2nd c. AD, and thus provide another reliable date for this process. In the same time, however, the Brahmins tried to gain ground again. Especially those among them who specialized in legal afffairs and political sciences and were closely related to the ruling elites began to adopt writing as one of their traditional disciplines of learning (kalā). It is possible to trace this gradual process in the literature of this time, e.g. the introduction of written documents into legal procedure as witnessed in the Dharmaśāstra literature (cf. Strauch 2002: 19–52). This development was accompanied by the growth of a new, mighty religious movement which is nowadays subsumed under the term Hinduism. The texts of this movement such as the great epos Mahābhārata with its influential Bhagavadgītā were no longer the secret knowledge of a small group of specialists but were widely propagated. Hinduism provided the ideological base for the spread of the Brahmanical culture over the entire Indian subcontinent (cf. now Malinar 2009: 50–66). The basic means of communication in this process was Sanskrit, which not only served as lingua franca, but also regained its status of a religiously legitimized literary language. It is signifijicant that the earliest Indian inscriptions which show clear traces of Sanskrit phonology are not originating from a Buddhist background, but can be ascribed to the newly arising Hinduist culture.12 The status of Sanskrit gradually became so strong and influential that even the Buddhists in India started to sanskritize their texts.13 Traces of this sanskritization are felt throughout: not only in early Buddhist literature but also in the inscriptions of that time (Damsteegt 1978). The Kharoṣṭhī epigraphs show that also the Gāndhārī speaking area was subject to this process from about the late 2nd/early 3rd c. AD onwards (Salomon 2001: 141). In India proper the specialists from the brāhmaṇa circles had developed the Brāhmī script into an adequate instrument for a phonologically correct reflection of the Sanskrit language. By the end of the 2nd century AD the Brāhmī disposed of a complete inventory of signs and sign combi-

12

 Cf. e.g. the Brāhmī epigraphs from Ghosuṇḍī and Hāṭhībāḍā (Salomon 1998a: 86f.).  Cf. for the general conditions of this process and the influence from the side of Brahmanical culture Salomon 2001: 248–251. Controversially discussed is the signifijicance of writing for the emergence of a new Buddhist movement, the Mahāyāna, which is said to have a special attachment to writing and written artifacts like books (for a summary of the recent discussion and further references cf. Drewes 2009). As recent research shows, Mahāyāna was also influential among Gandhāran Buddhism (Strauch 2010). 13

indian kharoṣṭhī script and the “sanskrit revolution”

151

nations in order to express Sanskrit in a non-ambiguous way. The general consistency of orthography and the graphical solutions which were found speak in favour of a somehow controlled or at least harmonized process based on a sound knowledge of Sanskrit phonology and grammar. The “Internal Sanskritization” of Kharoṣṭhī But how did the proponents of the Kharoṣṭhī script react to this Pan-Indian challenge? The number of manuscripts or inscriptions which belong to the group of sanskritized or Sanskrit texts is rather limited. Nearly all of them are quite late and are datable to the 3rd c. AD or even later, i.e. into the fijinal phase of Kharoṣṭhī (cf. Salomon 2001). Fortunately, among the texts of the Bajaur Collection there is one manuscript which seems to belong to an earlier phase of Kharoṣṭhī Sanskrit writing (= BajC 9 recto). According to its palaeographical features and the evidence of the collection as a whole it should not be later than the 2nd c. AD (Strauch 2008: 108–111). The manuscript contains a collection of verses which belong to the Brahmanical genre of Nīti literature, i.e. political science.14 This text shows that the process of sanskritization did not only involve a linguistic shift within the boundaries of Buddhist literature but did also include a cultural change which implied a more intensive confrontation with new branches of non-Buddhist literature composed in Sanskrit. The Bajaur manuscript BajC 9 is written in a conventional Kharoṣṭhī using the typical Arapacana signs and its modifijication markers. There is no extensive use of newly created conjunct signs or special indicators for vowel quantity which are typical for later varieties of Sanskrit in Kharoṣṭhī script. Therefore, at the fijirst glance the text looks like ordinary Gāndhārī. But phonetic features like the use of external and internal sandhi and morphological forms which are characteristic only for Sanskrit reveal its true language. Moreover, the text is composed in the Āryā meter. In most cases it is possible to reconstruct from the defective Kharoṣṭhī spelling the correct Sanskrit text. The principles which were used for writing Sanskrit can be demonstrated on the base of one of the verses which lists the components of the royal income (BajC 9 recto, verse 8):

14

 For more information about this text see Strauch 2008: 125–127.

152

ingo strauch

Fig. 1. Extract from the manuscript BajC 9, recto.

Kharoṣṭhī spelling

Sanskrit reconstruction

dhaṇa-dhanya-kupya-yavaseṃdhaṇ{e}ṇi yatrayudhani ca rathac̄a upakaraṇani ca kośo naravahana-śipi-yodhac̄a

dhana-dhānya-kupya-yavasendhanāni yantrāyudhāni ca rathāśca upakaraṇāni ca kośo naravāhana-śilpi-yodhāśca

Translation: The treasure (are) money, grain, forest produce,15 grass, machines and weapons, chariots and instruments, men, vehicles, craftsmen and soldiers. For giving a more complete picture of the overall appearance of the manuscript, the following survey will be complemented by selected evidence from other portions of the same text. 1. Vowel quantity (above = bold print) Throughout the manuscript the quantities of the vowels are not indicated. Other text portions show that also diphthongs are not indicated but represented by their respective monophthongs (ceva = Skt. caiva). The sonant r̥ is expressed either by ri (nripati = Skt. nr̥ pati) or ir (hirdaya = Skt. hr̥ daya), or—in a labial environment—by ur (vurdhi = Skt. vr̥ ddhi). 2. Simple consonants With the exception of the confusion of dental and retroflex nasal (n/ṇ), which is typical for most of the varieties of Kharoṣṭhī and reflects the supposed indistinctiveness of the pronunciation of both sounds all simple (short) consonants are represented by their expected Kharoṣṭhī equivalent (cf. above table 2). 15

 This connotation of kupya is obvious from the chapter 2.17 of the Kauṭilīya Arthaśāstra which is exclusively devoted to the duties of the “Director of forest produce” (kupyādhyakṣa). See also Kangle (1969: Glossary, s.v.).

indian kharoṣṭhī script and the “sanskrit revolution”

153

3. Consonant clusters (above = cursive print) Consonant clusters are indicated generally only in cases when the conventional inventory of Kharoṣṭhī offfers a possible grapheme. Geminated consonants are not indicated (t = Skt. tt). 3.1 Semivowel diacritics (-y, -r, r-, v-) Clusters containing the semivowels y and r are usually indicated by means of the primary modifijiers of the Kharoṣṭhī, like above nya, pya, tra. As an example for preconsonantal r atmartha-: Skt. ātmārtha- and durge: Skt. durge can be cited, postconsonantal v is found in kritva = Skt. kr̥ tvā. 3.2 Basic Arapacana signs expressing sound combinations Although the phonetic realization of these signs (Arapacana no. 25, 26, 38) is disputed, the scribe of BajC 9 apparently perceived them as representatives of Old Indian sound clusters. Thus we fijind in the text kṣa (no. 25) for Skt. kṣa (kośakṣaya = Skt. kośakṣaya), sta (no. 26) for Skt. sta (śastreṣu = Skt. śāstreṣu), and ḱa (no. 38) for Skt. ska (saṃḱrita = Skt. saṃskr ̥ta) (cf. above table 2). 3.3 Secondarily modifijied Arapacana signs In other cases the Kharoṣṭhī either uses the possibility of additional modifying signs, e.g. a bar above the letter (c̄a = śca) or an anusvāra indicating the class nasal (ṃdha = ndha). The horizontal bar is also used for the sound combination jña which is depicted as ña with bar above ( ), a sign which is so far unattested in Kharoṣṭhī palaeography. Another unattested combination is found in th̄a (

) which is used beside unmodifijied

tha (= Arapacana, no. 19, ) occasionally for Skt. stha. A diffferent internal derivation for this sound combination which is based on the Arapacana sign sta (no. 26) is found in later Central Asian Kharoṣṭhī ( , sthi, Glass 2000: 133). Another special modifijication is represented by clusters containing a pre- or postconsonantal l, where the vertical lower end of the basic sign is crossed by a stroke: klaṃ ( ), kle ( ), lkaṃ ( ). This rarely attested device can be associated with the ligature lpi found in the Kharoṣṭhī legends of Vima Kadphises ( ) and in later Central Asian Kharoṣṭhī ( (cf. Glass 2000: 132).

)

154

ingo strauch

3.4 No indication Occasionally consonant clusters are not indicated at all, even if the conventional Kharoṣṭhī would provide a solution (tr = ntr, p = lp). 3.5 Combined signs Only in very rare cases the scribe makes use of combined signs which join basic letters. Most of them are also attested in contemporary Kharoṣṭhī records, like e.g. tma (atmartha = Skt. ātmārtha). Furthermore, the Kharoṣṭhī of the Bajaur manuscript does not indicate Visarga or fijinal consonant. It is quite obvious that this Bajaur text represents an early attempt to cross the language border towards Sanskrit on the basis of the instrumentarium developed within the Kharoṣṭhī script. I would like to call this process “Internal Sanskritization of Kharoṣṭhī”. It has to be distinguished from the later phase which makes extensive use of the tools which have been developed and successfully used by the Brāhmī alphabet. Due to this indebtedness to an external source this later type can be named “External Sanskritization of Kharoṣṭhī”. The interaction of Brāhmī and Kharoṣṭhī While the influence of Brāhmī might have been rather weak in the Aśokan period16 it became more substantial in the subsequent centuries when Indo-Greek and Indo-Scythian rulers held sway over the North-West. The peak of this development was certainly reached under the dynasty of the Kuṣāṇas. From the end of the 1st c. AD onwards they started to establish an empire which united Bactria, the Hindukush area, the Indian North-West and the north of India up to Pāṭaliputra, i.e. modern Patna, thus covering a huge area where both Kharoṣṭhī and Brāhmī were in use. Although the Kuṣāṇas did not actively promote any of both scripts, the social and economic mobility within the boundaries of their empire and with its direct neighbours, the Western Kṣatrapas and Sātavāhanas in Gujarat and on the Deccan, resulted also in an interaction of both writing systems. We do not only fijind the occasional use of Brāhmī and Kharoṣṭhī in areas which were

16

 That biscriptuality was an old phenomenon, is shown by the case of the scribe Capaḍa who produced parts of the Aśokan edicts at Brahmagiri, Siddapur and Jaṭiṅga Rāmeśvara. He added to the otherwise Brāhmī texts his title lipikara “scribe” in Kharoṣṭhī letters (Salomon 1998: 136).

indian kharoṣṭhī script and the “sanskrit revolution”

155

originally remote to them17 but also the increasing occurrence of BrāhmīKharoṣṭhī biscriptual legends on seals and coins as well in Northern India as in the Kharoṣṭhī using area in the North-West. It might be interesting to notice that the Kuṣāṇas themselves did not actively participate in this process. Instead, under Kaniṣka I (after 127 AD) they replaced their originally biscriptual-bilingual Greek-Kharoṣṭhī legends, which they had inherited from their predecessors, by a single Bactrian legend. Other dynasties, however, as the neighbouring Kṣaharātas who shifted from the North-West to Gujarat, i.e. into a Brāhmī using area, continued the older system and added a Brāhmī legend, a tradition which was continued from Bhūmaka via Nahapāna up to the Kṣatrapa ruler Damazāda (Senior 2001: 194–200). The use of Kharoṣṭhī was abandoned when the Middle Indian legends were replaced by Sanskrit under the ruler Rudradāman. The opposite development is attested by the coinage of the Pāratarāja dynasty who ruled from the 2nd till the 4th c. AD in Baluchistan. The earliest rulers in the middle of the 2nd c. AD used exclusively Brāhmī legends before the dynasty shifted to the use of Kharoṣṭhī which can be safely dated up to the end of the 3rd c. AD.18 This evidence also shows the sphere of influence the Brāhmī reached during the fijirst centuries AD. Although it cannot be excluded that some examples of this early Kharoṣṭhī-Brāhmī biscriptuality were accompanied by bilinguality,19 most

17  For Kharoṣṭhī scribes in the Brāhmī area see e.g. the famous Mathurā Lion Capital inscription of the 1st c. AD (CKI 48, see now Falk 2011b) and the Mathurā bilingual inscription dated (Kuṣāṇa) year 40 = 167 AD (CKI 440, Chattopadhyaya 1980–82, Bhattacharya 1984). Even more eastern examples are found at Bharhut (Cunningham 1879: 8, pl. VIII) and Patna (CKI 166, Konow 1929: 177f., plate XXXVI). For a probably imported Kuṣāṇa period Brāhmī inscription found in the Peshawar valley see Falk 2004: 139f. Earlier examples for the use of Kharoṣṭhī outside its original territory are the 3rd–2nd c. BC biscriptual inscriptions from the Kangra valley in Himachal Pradesh (Vogel 1902–03, CKI 167+168). 18  For more information about the Pāratarājas and their coinage see Tandon 2006, 2009 and Falk 2007. The chronology used here follows Tandon 2009, the terminus post quem of the latest Kharoṣṭhī issues can be reliably fijixed on the base of overstruck coins (Tandon 2009: 154–156). Other contemporary biscriptual coinages like those of the Audumbaras (ca. 1st c. AD) and the Kuniṇḍas show that the introduction of Brāhmī cum Kharoṣṭhī legends was initially mainly restricted to territories and by dynasties which were peripheral to the core-land of Gandhāra and its ruling elites, but located at important trade routes which connected this area with Brāhmī using territories (cf. Chattopadhyaya 2003: 59–60).The much earlier attempt by Agathocles and Pantaleon (190–180 BC) to introduce Brāhmī on their coins is not considered here (cf. Bopearachchi 1991: 175, pl. 7, séries 9 +10, 182, pl. 9, série 6). It remained without further consequences for the development of writing in the region. 19  A clear case of bilinguality seems to be the Kharoṣṭhī-Brāhmī Kanhiāra inscription (Vogel 1902–03, Konow 1929: 178, CKI 168). Here the Sanskrit text kr ̥ṣṇayaśasya ārāma mādagisya written in Brāhmī is accompanied by the Gāndhārī kriṣ̄ayaśasa aramo in

156

ingo strauch

biscriptual epigraphs seem to represent the same Middle Indian language. If a more accurate defijinition of this language is possible at all, it can be attributed to the North-Western, i.e. Gāndhārī speaking region. Thus not only many of the biscriptual seals (e.g. Konow 1929: 100, 102) but also the biscriptual Kuṣāṇa period Mathurā epigraph (Bhattacharya 1984) are nothing more than Gāndhārī written in Brāhmī script. In many cases the scribes simply transferred Kharoṣṭḥī orthographical usages into the new script.20 This clearly shows the direction of assimilation and its agents. Obviously Kharoṣṭhī scribes tried to use Brāhmī for their own language, which might afffect the orthography of the Brāhmī written by them. As far as we can observe on the basis of the available evidence, there was no remarkable tendency to an opposite influence. It seems that as long as the media of communication was a Middle Indian language there was no need for a Kharoṣṭhī scribe to seriously revise his script and to leave the above described internal limits of its instrumentarium.21 With the beginning of the 3rd c. AD the picture changed. The Middle Indian language of Gandhāra had been reshaped into a kind of sanskritized lingua franca which was not very diffferent from the varieties of Sanskrit which were in use in other parts of the Indian subcontinent. This sanskritized language afffected all spheres of writing, i.e. not only epigraphical records but also the transmission of Buddhist literature. In the beginning

Kharoṣṭhī (my readings). The inscription which was found in the Kangra valley in modern Himachal Pradesh seems to be very early and can probably be dated into the 2nd c. BC. It supports the idea that the preferable medium for writing Sanskrit was Brāhmī while the use of Kharoṣṭhī was at that period confijined to Gāndhārī text. It is not clear, whether the bronze die with the text (Kharoṣṭhī) sidhatasa, (Brāhmī) śidhataśa is also such a case of bilinguality where the Brāhmī is meant to designate an East-Indian dialect—(Ardha-)māgadhī—which replaces dental s by palatal ś (Archaeological Survey of India, Annual Report 1936/37: 39, pl. X, f,g). It is also possible that it represents an example for the incorrect use of the sibilant which is typical for the earliest Brāhmī as used by Aśoka. See for this seal now Aman ur Rahman & Falk 2011: 186, TM 07.07.01. 20  A possible example for such a transfer of Kharoṣṭhī orthography is the use of preconsontal r to designate a geminated consonant in a Mathurā epigraph of the Kuṣāṇa period (Lüders 1961: 82f., § 46): derya = deyya. Such a usage of Kharoṣṭhī orthography can also be verifijied in much later documents. Thus we fijind in the frequently attested spelling seryathā = seyyathā in Buddhist Sanskrit texts (ibid.: 83) and in a peculiar Prātimokṣasūtra manuscript from Qizil of the 6th/7th c. (von Simson 1997) which reads e.g. adirna for adinna (583). For this device in Kharoṣṭhī texts see now Salomon 2008a: 97. 21  An interesting case how a “foreign” sound is designated in Kharoṣṭhī are the Pāratarāja names Kożana and Kożiya. The Kharoṣṭhī coins mentioning them introduced a new Kharoṣṭhī sign by modifying the letter z with the help of an additional hook added to its lower right and a circle on its top. The Brāhmī parallel seems to be based on the established ligature ysa /za/ which was, however, not used by the Kharoṣṭhī scribes as a model for their new sign (Falk 2007).

indian kharoṣṭhī script and the “sanskrit revolution”

157

of this sanskritization Kharoṣṭhī continued to be in use. In the sphere of Buddhist literature this is impressively proven by the sanskritized Buddhist texts in Kharoṣṭhī script which are part of the British Library, Pelliot and Schøyen Collections.22 According to recent radio-carbon dating the relevant Schøyen fragments can most probably be dated into the 3rd c. AD (Allon et al. 2006: 288–291). The inscriptional evidence of this period shows that the use of Kharoṣṭhī for this sanskritized Gāndhārī became gradually replaced by that of Brāhmī. In many cases the transition is connected with the advent of foreigners originating from Brāhmī using areas (Falk 2004). As far as we can judge from the available material, this process fijirst concentrated on two distinctive environments: A—Non-Buddhist establishments which obviously preferred the use of Brāhmī as a script which was not linked to the Buddhist culture of Gandhāra. As an example for this category the numerous epigraphs discovered in the early Hindu site Kashmir Smast can be cited. They show how the peculiar—still heavily Gāndhārī based—sanskritized language of the region was tried to be reproduced in the newly introduced Brāhmī script (cf. Falk 2001 and Falk 2008b: 138–143).23 B—Peripheral places of the Kharoṣṭhī using area with a high factor of social mobility which were less reluctant to accept a new script. The Buddhist monastery complex Kara-Tepe near Termez (Uzbekistan) can serve as a paradigmatic place for this category. Here we fijind the contemporary use of Brāhmī and Kharoṣṭhī probably datable into the 3rd–4th c. AD.24

22  The Sanskritic features of the respective manuscripts are described in Salomon 1998b (Pelliot), Salomon 1999: 123, Salomon 2001: 243 (British Library), Allon & Salomon 2000: 266–271, Salomon 2001: 243–247, and Allon et al. 2006: 288–290 (Schøyen). The same category of relatively late sanskritized Kharoṣṭhī texts is represented by a hitherto unpublished palm-leaf folio from the Kabul Museum which probably hails from a cave at Bamiyan (Falk & Strauch, forthcoming). 23  Forms such as the ending -sya for the genitive sg. masc. clearly show the Skt. character. The inconsistency in the designation of vowel quantity and prakritisms or hyper-sanskritisms such as pāṇīka and pāṇīā (for Skt. pāṇīya) and pratitthāvita (for Skt. pratiṣṭhāpita-) indicate the underlying Gāndhārī phonology. 24  The editor of these inscriptions suggests a much later date for some of the bilingual epigraphs comparing the records of the monk Buddhaśira(s) with the Brāhmī inscriptions of Central India and the Deccan of the 5th/6th c. AD (Vertogradova 1995: 32). On the base of the palaeographical arguments presented by her it seems, however, more plausible to presume a date in the late Kuṣāṇa/early Gupta period, i.e. around the end of the 3rd, beginning of the 4th c. AD. A still earlier date is now suggested by Gérard Fussman (2011:

158

ingo strauch

A series of biscriptual inscriptions on ceramic vessels25 shows that Brāhmī and Kharoṣṭhī virtually reproduced the same sanskritized language. That the gradual introduction of Brāhmī did also afffect places in the Gāndhārī core-land can be shown by a bilingual epigraph from a Buddhist site near Peshawar (Falk 2004: 146f., plate V, d) and by a vessel with a Brāhmī inscription from a Buddhist monastery near Haḍḍa called mahāsenāraṇya (pot 1: Sadakata 1999: 65–66, plate II). Interestingly, the same monastery name is attested on another ceramic pot inscribed in Kharoṣṭhī (pot 2: CKI 360, Salomon 1996: 238–242, plates 8–14). Here we read in sanskritized orthography: saghe cadurdiśe masenaranye26 “To the community of the four directions at the Mahāsena monastery.” Obviously the Mahāsenāraṇya has completed the shift from Kharoṣṭhī to Brāhmī in the period between the production of both vessels. It seems that this shift was preceded or accompanied by a change of Kharoṣṭhī orthography. The initial phase of this development can be illustrated with the help of another pair of pots. The same sanskritized spelling as found on the pot from the Ma(hā)senāraṇya is seen on a yet unpublished vessel from the Mahāpriyāraṇya monastery near Haḍḍa. Here the same unusual ligature ṇya and additionally even a long ā occur: mahāpriaraṇye (pot 3: Strauch, forthcoming). This evidence can be contrasted with the variant mahapriyaraṃñe in conventional Kharoṣṭhī orthography as found on another vessel from the same monastery (pot 4: CKI 223, Fussman 1969).27 All these vessel inscriptions represent diffferent stages in the development from conventional Kharoṣṭhī to Brāhmī, caused by the sanskritization of the North-Western language:

41–45) who dates the biscriptual texts of Buddhaśira(s) between 50 and 200 AD while Jīvananda is placed by him between 150 and 250 AD. 25  The multiscriptual inscriptions of the Kara-Tepe monks Jīvananda and Buddhaśira(s) are published by V.V. Vertogradova (1995: 106–113, 2004: 69–72) and Fussman 2011: 63–88 (Buddhaśira(s) alias Buddhamitra) and 67f. ( Jīvananda). Another triscriptual text of Jīvananda which was not recognized as such seems to be 51 KT (Fussman 2011: 75f.). The Kharoṣṭhī text on sherd 51 KT b can clearly be read as [j]i[va]na[ṃ]das[y]a. The script is identical with that used on the other pots of this person. 26  What Salomon interpretes as “long, decorative extensions to the last syllable of a word” (1996: 239) can now be safely identifijied as subscribed -ya. Hence his readings masenarane and budhaghoṣasa have to be corrected to masenaranye and budhaghoṣasya. For masena as contracted form of mahāsena cf. Salomon 1996: 242. 27  As Salomon (1999: 243) rightly points out, Fussman’s reading mahapriasaṃñe has to be corrected.

indian kharoṣṭhī script and the “sanskrit revolution” Language

Gāndhārī

Script

Conventional Kharoṣṭhī Modifijied Kharoṣṭhī Brāhmī

Mahāsenāraṇya Mahāpriyāraṇya mahapriyaraṃñe (pot 4)

159

Sanskritized language

masenaranye (pot 1) mahāpriaraṇye (pot 3)

mahāsenāraṇye (pot 2)

As this evidence shows, the process of sanskritizing did not only afffect the literary and administrative language of the region but also had a considerable impact on the orthography of the Kharoṣṭhī. As mentioned before, of both scripts it was the Brāhmī which had already found the basic answers to the challenge of sanskritization. What would be more natural than the attempt to use these answers for the adjustment of Kharoṣṭhī? The “External Sanskritization” of Kharoṣṭhī The following short survey will list the most characteristic changes of Kharoṣṭhī in the late phase of its adaption to Sanskrit phonology as witnessed by the few examples of sanskritized Kharoṣṭhī texts.28 Beside the few texts of the Schøyen, British Library and Pelliot collections the most extensive repertoire for this late type of Kharoṣṭhī is represented by the documents on wood and leather discovered in the beginning of the 20th c. by Aurel Stein in Niya and Endere (Boyer et al. 1920–29). Most of these documents are written in the administrative language of the Kroraina kingdom which is based on Gāndhārī but largely influenced by an underlying dialect related to Tokharian (cf. Burrow 1937). Some of them, however, contain Buddhist verses which are composed in the Buddhist Sanskrit of the period (Iwamatsu 1998, 2000, 2001, 2002). The entire corpus can be dated into the 3rd/4th c. AD (Brough 1965). Both the foreign elements of the language as well as the desire to express Sanskrit afffected the development of the Niya Kharoṣṭhī which is abundant in newly created sign modifijications and combinations (Boyer et al. 1920–29: 295–322). In the late Sanskrit or sanskritized Kharoṣṭhī texts consonant clusters are usually formed by writing the respective basic signs one above the

28

 Since the number of these texts can be expected to increase and only some of the known texts are published so far, the survey is just a cursory overview to indicate the general characteristics. The major texts which are available at present are listed in Salomon 2001 and Allon et al. 2006.

160

ingo strauch

other. This method clearly adopts the way which is the standard one for Brāhmī. As opposed the “internal” way of expressing such combinations by non-standardized diacritics, this approach guarantees a phonologically adequate reading.29 The necessary unambiguity is, however, never reached, since all Kharoṣṭhī texts attested so far simultaneously continue the less distinct “internal method”. Thus despite its otherwise sanskritized or—better—brahmiized character the Schøyen manuscript applies s̄ for Skt. sna and j̄ for Skt. jha (Allon & Salomon 2000: 267). Table 4.30 Selected consonant clusters in late Kharoṣṭhī as opposed to the orthography of the manuscript BajC 9 External Sanskritization Skt.

Schøyen Collection

Pelliot Collection

Niya documents

Internal Sanskritization Bajaur fragment 9

kta

ta

jña

ña

tta

ta

ṣṭha

ṭha

stha

th̄a

lpa

lkaṃ

29

 An interesting and rather early example for this device is found with the ligature

ṣthu ( ) on the two Wardak vase inscriptions dated into the (Kuṣāṇa) year 51 (= 178 AD) (CKI 159 = Konow 1929: 165–179, CKI 509 = Falk 2008a). Remarkably, it occurs only in the foreign, probably Iranian, name Haṣthunä. The two dots above the fijinal letter (nä) seem to indicate another orthographical peculiarity which was probably introduced into Kharoṣṭhī from a foreign writing system. Konow (1929: 166) compares this device with the vowel designation -ä of Śaka Brāhmī texts. Usually such double dot above a consonant sign is used to designate the Skt. Visarga and probably derived from the corresponding Brāhmī sign. Cf. also the discussion in Salomon 1998: 131, 143 and Glass 2000: 137f. 30  With the exception of stha (Allon & Salomon 2000: pl. X,1) and the BajC signs, all Kharoṣṭhī signs are extracted from Glass 2000.

indian kharoṣṭhī script and the “sanskrit revolution”

161

Table 5.31 Additional signs of late Kharoṣṭhī and their Brāhmī models Vowel quantity

Visarga

Virāma

Brāhmī

ā

yaḥ

tam

Kharoṣṭhī

ā

taḥ

dhik

Other orthographical needs for writing Sanskrit were only sporadically satisfijied. Thus the Visarga is found only in the few Kharoṣṭhī manuscripts from the Northern Silk Road which are now part of the Pelliot Collection (Salomon 1998b) and in few of the later Kharoṣṭhī tablets from Central Asia (cf. ibid.: 131). In Kuṣāṇa times the Brāhmī had started to write this sound as double dot on the right side of a letter. The new Kharoṣṭhī sign which arranges these dots on the top of the basic sign can be interpreted as an adaptation of this method. Its occurrence in Central Asia is clearly based on the prominent presence of Brāhmī in this area. Another feature which was introduced rather late into Kharoṣṭhī is the indication of vowel quantity. The earliest attempt seems to be represented in the Avadāna texts of the British Library Collection scribe 2 from about the 2nd century which is restricted, however, to the initial ā (Lenz 2010: 17). The long quality of the vowel is indicated by a stroke to the lower right of the letter which can be associated to the shape of the -ā diacritic as found in some varieties of Kuṣāṇa Brāhmī (cf. table 5). The same practice is also occasionally attested in epigraphical records, generally in the case of inscriptions which show Sanskrit influence and were written in a biscriptual environment (e.g. Kara-Tepe, Vertogradova 1995: 19f.). It remained in use in Central Asia where we fijind the only text of Kharoṣṭhī literature which shows a consequent approach to the indication of vowel quantity (Niya document no. 523, Boyer et al. 1920–29: 191, tr. Burrow 1937: 103). Remarkably, this tablet contains also one of the few Kharoṣṭhī evidences for the indication of fijinal consonants. In Kuṣāṇa Brāhmī this problem was solved by adding a consonant sign in a smaller size than the preceding. The same device was used by the scribe of Niya document 523, obviously under the influence of the Brāhmī orthography (cf. Glass 2000: 138). This

31

 The Kharoṣṭhī signs are extracted from Glass 2000, the Kuṣāṇa Brāhmī signs ā and yaḥ from the Indoskript database (http://userpage.fu-berlin.de/~falk/), the tam from Sander 1968: table II.

162

ingo strauch

influence is also traceable in the shape of the numerals 1–3 written as horizontal strokes instead of the usual verticals. The same brahmiized numerals are found in some of the Schøyen manuscripts (Glass 2000: 139f.). These few observations might be sufffijicient to characterize the main features of the external sanskritization of Kharoṣṭhī by the end of the 3rd c. AD. The technical realization of Sanskrit phonology as well as the contexts of the inscriptions and manuscripts indicate that this process took place in a distinctively biscriptual Brāhmī-Kharoṣṭhī environment where the Kharoṣṭhī was exposed to the principles of Brāhmī orthography and tried to adapt them for the reproduction of the newly introduced Sanskrit language. The influence of Brāhmī on the Kharoṣṭhī orthography, however, did never result in a consistent adjustment of Kharoṣṭhī to the Sanskrit phonology. Besides adapting Brāhmī orthographical devices the Kharoṣṭhī scribes continued to use a multitude of diacritical modifying signs and never tried to reduce the inventory of letters down to a reasonable number. Conclusions Kharoṣṭhī was designed for the specifijic needs of the North-Western language Gāndhārī and remained closely linked to this region and language throughout its use. As shown by its own inventory of signs—as represented in the Arapacana alphabet—it is a distinctly phonetic script which tries to reproduce phonetic variants by a multitude of basic signs and supplementary modifijiers. An attempt has never been made to transform the script into a phonological writing system which confijines itself to the consequent tagging of phonemes. Moreover, the phonetic character of the script was never regulated by commonly accepted orthographical standards. The attested attempts to use Kharoṣṭhī for writing Sanskrit or a sanskritized language can be attributed to two diffferent and clearly distinguishable approaches. While the fijirst one—which I call “Internal Sanskritization”— uses the instruments developed within the limits of Kharoṣṭhī writing by applying the basic Arapacana signs and their primary and secondary modifijications, the second approach is clearly influenced by the methods which are peculiar for the Brāhmī. This “External Sanskritization”, which could also be labeled as “Brahmiization”, is characterized by the increasing use of combined signs (ligatures) and graphical devices which are otherwise confijined to the Brāhmī script (Visarga, Virāma). Although the “internally sanskritized” Kharoṣṭhī seems to be historically older, it did not

indian kharoṣṭhī script and the “sanskrit revolution”

163

become obsolete after the application of the external (Brāhmī) devices. Both methods continued to be used within the same text up to the very end of the Kharoṣṭhī period. To enable Kharoṣṭhī to cross its linguistic limits and to be used for Sanskrit would have required a serious reform of the writing system which would have consequently adjusted its inventory of signs to the needs of Sanskrit phonology. Such a reform would have presupposed a strong cultural or political agent with a profound interest in maintaining Kharoṣṭhī as a writing system. It seems that this interest did not arise in Gandhāra proper when by the end of the 3rd c. AD the use of Gāndhārī as lingua franca and literary language came to an end and had to give way to Sanskrit. Neither the political elites nor the Buddhist institutions resisted when this linguistic shift was accompanied by the introduction of a new script which was much better adapted to Sanskrit phonology and in this regard much superior to Kharoṣṭhī. Although the Buddhists had made a certain attempt to sacralize the Kharoṣṭhī script by ascribing the Arapacana alphabet a magic and ritual power, this did not result in a status which would have prevented their shift to the new alphabet. The diffferent versions of the Lalitavistara story of the young Bodhisattva as a school-boy show impressively how pragmatically the Buddhists handled this problem. In this story the future Buddha is supposed to learn the alphabet. Of course, due to his outstanding capacities his knowledge of scripts supersedes that of his teacher by many times. Altogether 64 diffferent scripts are enumerated by the future Buddha, the fijirst of them being Brāhmī and Kharoṣṭhī. Moreover, when the school-master starts to teach the alphabet, the present young boys and the Bodhisattva do not only learn the respective letters but also a great number of religious concepts which are associated to each of them. While in the older version, preserved in Dharmarakṣa’s Chinese translation of this text (308 AD), it is the Arapacana order which the teacher is reciting (Brough 1977), its younger Sanskrit parallel rearranges the whole passage according to the sequence of the Sanskrit varṇamālā. The continuing use of Kharoṣṭhī in some other areas was mainly caused by the status the Gāndhārī language pertained in these peripheral communities. As long as Gāndhārī or a related dialect was in use as administrative or literary language, Kharoṣṭhī continued to be used as well. When the Gāndhārī lost this status and began to be replaced by Sanskrit the use of Kharoṣṭhī came to an end as well. The gradual displacement of Kharoṣṭhī was consequently the direct result of the gradual sanskritization of its cultural environment.

164

ingo strauch

Accordingly, the mysterious decline of Kharoṣṭhī is not so much due to a dynastical or political change, as assumed by Richard Salomon,32 but to a cultural and linguistic shift which is known under the term sanskritization. The new language brought along a well-adjusted writing system which was readily adapted in the writing culture of the area. As shown above, this pragmatical shift from Kharoṣṭhī to Brāhmī had both internal and external reasons. While the phonetic character of the Kharoṣṭhī script and its missing orthographical standardization were serious internal barriers, it was mainly the existence of the superior Brāhmī alphabet which externally prevented the Kharoṣṭhī to cross the border which was erected by the advent of the Sanskrit culture. Acknowledgements I am very grateful to Harry Falk with whom I discussed many aspects of the problems raised in this paper. He also read the fijinal draft and made a series of valuable suggestions. The Kharoṣṭhī signs from manuscripts of the Bajaur Collection were extracted by Andrea Schlosser. Many thanks go to my friend Gordon Adie and to Heidrun Wenisch, who checked the English of one of the “fijinal” versions. All remaining mistakes are, of course, exclusively my fault. References Allon, Mark 2007. The Senior manuscripts. In Andrew Glass: Four Gāndhārī Saṃyuktāgama Sūtras: Senior Kharoṣṭhī Fragment 5. Gandhāran Buddhist texts 4, Seattle: University of Washington Press: 3–25. Allon, Mark & Richard Salomon 2000. Kharoṣṭhī fragments of a Gāndhārī version of the Mahāparinirvāṇasūtra. In Jens Braarvig (ed.): Buddhist Manuscripts, volume I. Manuscripts in the Schøyen Collection I. Oslo: Hermes Academic Publishing: 243–73. Allon, M., Salomon, R. Jacobsen, G. & U. Zoppi 2006. Radiocarbon dating of Kharoṣṭhī fragments from the Schøyen and Senior manuscript collections. In Jens Braarvig (ed.): Buddhist Manuscripts, volume III. Manuscripts in the Schøyen Collection. Oslo: Hermes Academic Publishing: 279–91. Aman ur Rahman & Harry Falk 2011. Seals, sealings and tokens from Gandhāra. Monographien zur indischen Archäologie, Kunst und Philologie, Band 20 = Studies in the Aman ur Rahman Collection, 1. Wiesbaden: Reichert Verlag.

32

 It is hardly possible to qualify Kharoṣṭhī as “bureaucratic script of the Kuṣāṇa empire” (Salomon 2008b: 149). The only “offfijicial” epigraphical records of this dynasty are written in Bactrian, which was certainly the preferred language. What we know of the language policy seems to indicate that the language and script of the respective region was used in administrative afffairs.

indian kharoṣṭhī script and the “sanskrit revolution”

165

Baums, Stefan 2009. A Gāndhārī Commentary on Early Buddhist Verses: British Library Kharoṣṭhī Fragments 7, 9, 13 and 18. PhD Dissertation. University of Washington. Bhattacharya, Gouriswar 1984. On the fragmentary, bi‐scriptual pedestal inscription from Mathurā. Indian Museum Bulletin 19: 27–30. Bopearachchi, Osmund 1991. Monnaies gréco-bactriennes et indo-grecques. Catalogue raisonné. Paris: Bibliothèque Nationale. Boyer, A.M., E.J. Rapson, E. Senart, & J. Noble 1920–9. Kharoṣṭhī Inscriptions discovered by Sir Aurel Stein in Chinese Turkestan. 3 parts (part 3 by E. J. Rapson and P. S. Noble). Oxford: Clarendon Press. Brough, John 1962. The Gāndhārī Dharmapada. London Oriental Series 7. London: Oxford University Press. —— 1965. Comments on third-century Shan-shan and the history of Buddhism. Bulletin of the School of Oriental and African Studies 28: 582–612. —— 1977. The Arapacana syllabary in the old Lalita‐vistara. Bulletin of the School of Oriental and African Studies 40: 85–95. Burrow, Thomas 1937. The Language of the Kharoṣṭhi Documents from Chinese Turkestan. Cambridge: Cambridge University Press. —— 1940. A Translation of the Kharoṣṭhi Documents from Chinese Turkestan. James G. Forlong Fund 20. London: Royal Asiatic Society. Chattopadhyaya, B.D. 1980–2. On a bi‐scriptural epigraph of the Kuṣāṇa period from Mathurā (with a note on eghaṭṭa and ehaḍa by K. Meenakshi). Journal of Ancient Indian History 13: 277–84. —— 2003. Studying Early India: Archaeology, Texts and Historical Issues. Delhi. Coningham, R.A.E., Allchin, R.R., Batt C.M. & D. Lucy. 1996. Passage to India? Anuradhapura and the Early Use of the Brahmi Script. Cambridge Archaeological Journal 6: 73–97. Cunningham, Alexander 1879. The Stûpa of Bharhut. London: W.H. Allen and Co. Damsteegt, Th. 1978. Epigraphical Hybrid Sanskrit: Its Rise, Spread, Characteristics and Relationship to Buddhist Hybrid Sanskrit. Orientalia Rheno‐Traiectina, volumen vicesimum tertium. Leiden: E. J. Brill. Dani, Ahmad Hasan 1963. Indian palaeography. Oxford: Clarendon Press. Daniels, Peter T. & William Bright (eds.): The World’s Writing Systems. New York: Oxford University Press. Drewes, David 2009. Early Indian Mahāyāna Buddhism I: Recent Scholarship. Religion Compass 4, 2: 55–65. (http://www3.interscience.wiley.com/journal/117982875/home, accessed 8.3.2010). Falk, Harry 1993. Schrift im alten Indien: ein Forschungsbericht mit Anmerkungen. ScriptOralia, 56. Tübingen: Gunter Narr Verlag. —— 1996. Aramaic script and the Kharoṣṭhī—a comparison. Berliner Indologische Studien 9–10: 151–6. —— 2003. A copper plate donation record and some seals from the Kashmir Smast. Beiträge zur allgemeinen und vergleichenden Archäologie 23: 1–19. —— 2004. Six early Brāhmī inscriptions from Gandhāra. Università degli Studi di Napoli “L’Orientale”, Annali 64: 139–55. —— 2006. Aśokan Sites and Artefacts: a Source‐Book with Bibliography. Monographien zur indischen Archäologie, Kunst und Philologie, Band 18. Mainz am Rhein: Verlag Philipp von Zabern. —— 2007. The names of the Pāratarājas issuing coins with Kharoṣṭhī legends. The Numismatic Chronicle: 171–8. —— 2008a. Another reliquary vase from Wardak and consecrating fijire rites in Gandhāra. In Claudine Bautze‐Picron (ed.): Religion and art: new issues in Indian iconography and iconology. London: The British Association for South Asian Studies: 63–80. —— 2008b. Money can buy me heaven: religious donations in late and post‐Kushan India. Archäologische Mitteilungen aus Iran und Turan 40: 137–48.

166

ingo strauch

—— 2011a. The ‘Split’ Collection of Kharoṣṭhī texts. Annual Report of The International Research Institute for Advanced Buddhology at Soka University for the Academic Year 2010 [= ARIRIAB] 14: 13–23. —— 2011b. Ten thoughts on the Mathura Lion capital reliquary. In Shailendra Bhandare & Sanjay Garg (eds.). Felicitas: Essays in Numismatics, Epigraphy and History in Honour of Joe Cribb. Mumbai: Reesha Books International: 121–141. Falk, Harry & Ingo Strauch. Forthcoming. The Bajaur and Split Collections of Kharoṣṭhī manuscripts within the context of Buddhist Gāndhārī literature. In Paul Harrison & Jens-Uwe Hartmann (eds.): Proceedings of the conference “Indic Buddhist Manuscripts: The State of the Field”, Stanford, June 15–19, 2009. Fussman, Gérard 1969. Une inscription kharoṣṭhī à Haḍḍa. Bulletin de l’École française d’Extrême‐Orient 56: 5–9. —— 2011. Monuments bouddhiques de Termez. Termez Buddhist Monuments. I. Catalogue des inscriptions sur poteries (avec une contribution de Nicholas Sims-Williams et la collaboration d’Éric Ollivier). Collège de France. Publications de l’Institut de civilisation indienne, 79. Paris: Difffusion de Boccard. Glass, Andrew 2000. A Preliminary Study of Kharoṣṭhī Manuscript Paleography. MA thesis. Department of Asian Languages and Literature, University of Washington. —— 2007. Four Gāndhārī Saṃyuktāgama Sūtras: Senior Kharoṣṭhī Fragment 5. Gandhāran Buddhist Texts, 4. Seattle: University of Washington Press. von Hinüber, Oskar 1990. Der Beginn der Schrift und frühe Schriftlichkeit in Indien. Abhandlungen der geistes‐ und sozialwissenschaftlichen Klasse, Jahrgang 1989, Nr. 11. Mainz: Akademie der Wissenschaften und der Literatur. Iwamatsu, Asao (岩松浅夫) 1998. Karōshutī bunjochū no Bukkyō bongo no geju ni tsuite (1) カローシュチイー文書中の仏教梵語の偈頌について (1) (On some Buddhist Hybrid Sanskrit verses in the Kharoṣṭhī inscriptions mainly discovered by Sir Aurel Stein in Chinese Turkestan). Jinbun‐Ronshū 人文論集 (Studies in Humanities) 10: 1–16. —— 2000. Karōshutī bunjochū no Bukkyō bongo no geju ni tsuite (2) カローシュチイ ー文書中の仏教梵語の偈頌について (2) (On some Buddhist Hybrid Sanskrit verses in the Kharoṣṭhī inscriptions mainly discovered by Sir Aurel Stein in Chinese Turkestan). Siruku Rōdo kenkyū シルクロード研究センター (Silk Road Studies) 2: 21–35. —— 2001. Karōshutī bunjo daiban 511 ni tsuite (1) カローシュチイー文書第番511に ついて(1) (On the no. 511 text of the Kharoṣṭhī inscriptions). Jinbun‐Ronshū 人文論集 (Studies in Humanities) 13: 157–91. —— 2002. “Karōshutī bunjo daiban 511 ni tsuite (2) カローシュチイー文書第番511に ついて (2) (On the no. 511 text of the Kharoṣṭhī inscriptions). Jinbun‐Ronshū 人文論集 (Studies in Humanities) 14: 183–204. Kangle, R.P. (ed.) 1969. The Kauṭilīya Arthaśāstra. Part 1. A critical edition with a glossary. Bombay. Koizumi, Yoshihide (小泉 惠英) 2007. “Zāru Derī iseki shutsudo ishi chōgun no fukugenteki kōsatsu ザールデリー遺跡出土石彫群の復元的考察.” MUSEUM: Tōkyō kokuritsu hakubutsukan kenkyū shi MUSEUM: 東京国立博物館研究誌 606: 5–22. Konow, Sten 1929. Kharoshṭhī Inscriptions with the Exception of Those of Aśoka. Corpus Inscriptionum Indicarum 2.1. Calcutta: Government of India, Central Publication Branch. Lenz, Timothy 2010. Gandhāran Avadānas: British Library Kharoṣṭhī Fragments 1–3 and 21 and Supplementary Fragments A–C. Gandhāran Buddhist texts 6. Seattle: University of Washington Press. Lüders, Heinrich 1961. Mathurā inscriptions. Unpublished papers. Abhandlungen der Akademie der Wissenschaften in Göttingen. Phil.-hist. Kl. Dritte Folge 47. Göttingen. Malinar, Angelika 2009. Hinduismus. Studium Religionen. Göttingen: Vandenhoeck & Ruprecht. Pagel, Ulrich 2007. Mapping the Path: Vajrapadas in Mahāyāna Literature. Studia Philologica Buddhica. Monograph Series, XXI. Tokyo: The International Institute for Buddhist Studies of The International College for Postgraduate Buddhist Studies.

indian kharoṣṭhī script and the “sanskrit revolution”

167

Sadakata, Akira 1999. Girugitto shutsudo oyobi Bāmiyan shutsudo no Bukkyō kankei no Moji shinyō (Buddhist Manuscripts and Inscriptions Found in Gilgit and in Bamiyan). Tōkai Daigaku Kiyō Bungakubu (Bulletin of the Faculty of Letters, Tokai University) 71: 55–74. Salomon, Richard 1990. New evidence for a Gāndhārī origin of the Arapacana syllabary. Journal of the American Oriental Society 110: 255–73. —— 1996. Five Kharoṣṭhī inscriptions. Bulletin of the Asia Institute 10: 233–46. —— 1998a. Indian Epigraphy: A Guide to the Study of Inscriptions in Sanskrit, Prakrit, and the Other Indo‐Aryan Languages. South Asia Research. New York: Oxford University Press. —— 1998b. Kharoṣṭhī manuscript fragments in the Pelliot collection, Bibliothèque nationale de France. Bulletin d’études indiennes 16: 123–60. —— 1999. Ancient Buddhist Scrolls from Gandhāra: the British Library Kharoṣṭhī Fragments. Seattle: University of Washington Press. —— 2001. ‘Gāndhārī Hybrid Sanskrit’: new sources for the study of the Sanskritization of Buddhist literature. Indo‐Iranian Journal 44: 241–52. —— 2004. An Arapacana abecedary from Kara Tepe (Termez, Uzbekistan). Bulletin of the Asia Institute 18: 43–51. —— 2006. Kharoṣṭhī syllables used as location markers in Gandhāran stūpa architecture. In Pierfrancesco Callieri (ed.): Architetti, capomastri, artigiani: l’organizzazione dei cantieri e della produzione artistica nell’Asia ellenistica: studi offferti a Domenico Faccenna nel suo ottantesimo compleanno. Serie orientale Roma, vol. C. Roma: Istituto Italiano per l’Africa e l’Oriente: 181–224. —— 2008a. Two Gāndhārī Manuscripts of the Songs of Lake Anavatapta (Anavatapta‐gāthā): British Library Kharoṣṭhī Fragment 1 and Senior Scroll 14. Gandhāran Buddhist Texts, 5. Seattle: University of Washington Press. —— 2008b. Whatever happened to Kharoṣṭhī? The fate of a forgotten Indic script. In John Baines, John Bennet and Stephen Houston (eds.): The Disappearance of Writing Systems: Perspectives on Literacy and Communication. London: Equinox Publishing: 139–55. Sander, Lore 1968. Paläographisches zu den Sanskrithandschriften der Berliner Turfansammlung. Verzeichnis der Orientalischen Handschriften in Deutschland, Supplementband 8. Wiesbaden: Steiner. Senior, Robert 2001. Indo-Scythian Coins and History. Volume II. Lancaster, Pennsylvania: Classical Numismatic Group. von Simson, Georg 1997. Eine Prātimokṣasūtra-Handschrift in hybrider Sprache. In Petra Kiefffer-Pülz & Jens-Uwe Hartmann (eds.): Bauddhavidyāsudhākaraḥ. Studies in Honour of Heinz Bechert. On the Occasion of His 65th Birthday. Indica et Tibetica 30, SwisttalOdendorf: Indica et Tibetica Verlag: 583–604. Strauch, Ingo 2002. Die Lekhapaddhati-Lekhapañcāśikā. Briefe und Urkunden im mittelalterlichen Gujarat. Text, Übersetzung, Kommentar, Glossar. Monographien zur Indischen Archäologie, Kunst und Philologie, Band 16. Berlin: Dietrich Reimer Verlag. —— 2005. Urbanisierung, Antiurbanismus und Deurbanisierung: Wege zur Stadt im alten Indien. In Harry Falk (ed.): Wege zur Stadt. Entwicklung und Formen urbanen Lebens in der alten Welt. Vergleichende Studien zu Antike und Orient 2. Bremen: Ute Hempen Verlag: 121–157. —— 2008. The Bajaur collection of Kharoṣṭhī manuscripts—a preliminary survey. Studien zur Indologie und Iranistik 25: 103–36. —— 2010. More missing pieces of Early Pure Land Buddhism: New evidence for Akṣobhya and Abhirati in an early Mahayana sūtra from Gandhāra. The Eastern Buddhist 41: 23–66. —— forthcoming. Inscribed vessels and the geography of ancient Haḍḍa (working title). Tandon, Pankaj 2006. New light on the Pāratarājas. Numismatic Chronicle 166: 173–209. —— 2009. Further Light on the Pāratarājas: an Absolute Chronology of the Brāhmī and Kharoṣṭhī Series. Numismatic Chronicle 168: 137–171.

168

ingo strauch

Vertogradova, V.V. 1995. Indiĭskai͡a ėpigrafijika iz Kara-tepe v Starom Temeze: problemy deshifrovki i interpretat͡ sii. Индийская эпиграфика из Кара-тепе в Старом Термезе: проблемы дешифровки и интерпретации. Moskva: Izdatelʹskai͡a fijirma “Vostochnai͡a literatura” RAN. Москва Издательская фирма ”Восточная литература” РАН. —— 2004. Indian epigraphy and inscriptions in unknown writing from ancient Termez (Kara‐tepe, 1990–1994). International Association for the Study of the Cultures of Central Asia, Information Bulletin 24: 50–85. Vogel, J. Ph. 1902–03. Two Brahmi and Kharoshthi Rock‐Inscriptions in the Kangra Valley. Epigraphia Indica 7: 116–119. Voigt, Rainer 2005. Die Entwicklung der aramäischen zur Kharoṣṭhī- und Brāhmī-Schrift. Zeitschrift der Deutschen Morgenländischen Gesellschaft 155: 25–50.

SYMMETRY AND ASYMMETRY CHINESE WRITING IN JAPAN: THE CASE OF KOJIKI (712) Aldo Tollini Writing in Japan was imported from China. The Japanese started to write when they imported from the continent Chinese culture, considered superior and which included among other items, Buddhism and written texts. Therefore, the Japanese learned the existence of writing and its function only around the fijifth or sixth century AD when intensive exchanges across the channel that divides China from Japan started. In the beginning, the Japanese introduced written texts to their own country, most of which were Confucian or Buddhist in content. They tried to understand them asking the help of Korean or Chinese immigrants who worked as teachers and translators. In this way, they discovered the function and advantages of writing and they, subsequently, started to write. The Japanese originally imitated Chinese texts and wrote in Chinese, but little by little they started to use Chinese characters—called kanji 漢字—also to represent their autochthonous language, that is Japanese, often producing a hybrid language, not completely Chinese, nor completely Japanese. We have many texts dating to around the seventh and eighth centuries AD where kanji are employed to write Chinese, or Japanese, or even such a hybrid language. It was difffijicult to write an agglutinative language as Japanese with sufffijixes, inflections and functional words by means of characters born in an isolating linguistic environment. Many diffferent strategies were devised for that purpose before reaching the logophonographic writing system that is used up to the present day. At the beginning of the tenth century, the two parallel and complete phonetic syllabaries called hiragana and katakana that had been devised from kanji, started to be used largely in prose and poetry. They were used both to write the entire pure Japanese autochthonous language and to represent the functional parts of a more sinicized language. In that period, the Japanese used to write using diffferent styles called buntai 文体, which ranged from pure Japanese to pure Chinese, incuding also mixed strategies which can be called sino-japanese, and they were widespread. In principle, pure Japanese was written entirely (or mostly) phonographically with kana ( generally hiragana), while pure Chinese was written only in kanji. The hybrid

170

aldo tollini

forms of the written language used both kinds of characters, normally entrusting the semantic part to kanji and the functional parts to kana, which, being phonographic, could better express inflections and particles. Therefore, from around the tenth century onwards, we have texts written entirely in kanji, others written entirely in hiragana, or with a mixed form of writing that is kanji and kana called kanji kana majiri or kanji mixed with kana, that is the forerunner of the present day form of writing the Japanese language. In this way, kana greatly helped to make the writing of Japanese easier and more widespread. It is clear that Japanese writing from the beginning up to this day has always maintained a strong influence of Chinese writing and language. It is then interesting to investigate the symmetries and asymmetries that developed between the Chinese writing stystem and the Japanese one, which derived from the former. In fact, this is one of the few cases in which a so-called “logographic writing” (Chinese) developed into a mixed logophonographic one, maintaining many logographic traits, but integrating it with the necessary phonographic elements. The whole history of Japanese writing is closely connected to Chinese characters and also to the Chinese language, in a sort of unavoidable fate that accompanied its birth and development. In principle, the adaptation of Chinese characters in Japan used the following strategies, notwithstanding certain exceptions: 1. Chinese character/words imported as such and read with sino-japanese pronunciation, called on (音).1 In this case, the use of characters is logographic. Same character and same meaning as Chinese, and approximate Chinese reading. 2. Chinese character/words were given Japanese values, that is, the Japanese translated the Chinese character/words into their own language (this is called the wakun 和訓 strategy). In this case also, the use of characters is logographic. Same character and same meaning as Chinese, but use of a Japanese word, called kun (訓). 3. Chinese characters used only for their phonetic value to represent syllables of the Japanese language. In this case, the use of characters is phonographic. This case is doublefold: the reading of the character could be derived from the sino-japanese or the autochthonous reading, 1  Generally speaking, the sino-japanese pronunciation is the adaptation of the Chinese reading of characters to the Japanese phonological system (which was much simpler and regular compared to the Chinese one). It is an approximate Chinese reading.

symmetry and asymmetry

171

that is on or kun could be used. The fijirst case is called shakuon 借音, i.e., “loan of the on reading”, or ongana 音仮名, i.e., “kana read in on”. The second case is called shakukun 借訓, i.e., “loan of the kun reading”, or kungana 訓仮名, i.e., “kana read in kun.” For example, in the poetic anthology of Man’yōshū (middle of the 8th century) we fijind: 1. 山跡 for Yamato (the ancient name of Japan), and 2. 佐久良 for sakura, the cherry blossoms. The fijirst is a shakukun using autochthonous words: 山: yama “mountain”, 跡: (a)to, “footprint”. The second is a shakuon using a sino-japanese pronunciation of Chinese character/words: 佐: sa, “help”, 久: ku “long period of time”, 良: ra “good”. In both cases the meaning is neglected and the characters are used only for their phonetic value. In sum, in ancient Japan both the logographic and the phonographic strategies were present and often they were used together in various ways. Hereafter, I will try to present the interaction between written Japanese and Chinese from the lexical point of view, highlighting the difffijiculties and peculiarities born from this asymmetrical symmetry, if I can say so. The key point is the employment in Japan of Chinese logographic characters (symmetry) used in an isolating language as Chinese for an agglutinative language as Japanese (asymmetry). In order to discuss the matter concretely, I will take the example of Kojiki 古事記 (712 AD) (“Chronicle of Ancient Events”) that has two important characteristics that will help to illustrate the above subject: it is the fijirst attempt of extended writing in Japan (before it, only short, or very short pieces of writing were produced), and as declared in the preface in Chinese language, Kojiki is the written version of an oral tradition regarding ancient events, transmitted from generation to generation. For these two reasons Kojiki is particularly important for the development of writing in Japan. In writing Kojiki, the past examples of written texts were used, but in many cases new writing strategies were devised in order to write an extended text. For the fijirst time, in fact, an oral tradition transmitted from generation to generation in the autochthonous language had to be recorded by means of written characters. We have a unique chance to witness the birth of a written language (and the graphic representation of the oral autochthonous language). In the preface of Kojiki we fijind the fijirst important linguistic consideration about the difffijiculty of writing the Japanese language with Chinese characters:

172

aldo tollini In the ancient times, words and their meanings were simple, and therefore it is difffijicult (now) to write ( Japanese language) with Chinese characters. If kun are used, words cannot give the full sense, while using only on, the sense of what is written becomes long (prolix). Therefore, in this text, in the same sentence, I use kun and on together, or otherwise (were possible) only kun is used.

At those times, in Japan the model for writing, or the model of written language (which may not coincide), was Chinese language. However, Chinese language, called kanbun (漢文), was a very diffferent language from Japanese, and quite a lot of problems arose when trying to write Japanese with Chinese characters. One of which was that in writing Japanese, the prestigious model of Chinese language was adopted as far as possible giving the text a Chinese flavour and in the meantime making reading a complicated task. The actual written text of Kojiki is now considered by the modern linguists a kind of hybrid language—not Chinese, but not even fully Japanese—to which they gave the name of hentai kanbun (変体漢文) or “modifijied Chinese”: from an oral text OT1, Ō no Yasumaro, the compiler, wrote a hybrid text, from which the readers reproduced an oral text OT2, possibly the same as OT1. This means a transfer from a synthetic oral language OT1 ( Japanese),2 to an analytic written language (hentai kanbun),3 and fijinally again to a synthetic oral language OT2, as shown in the following scheme: OT1 (synthetic) ⇒ hentai kanbun (analytic) ⇒ OT2 (synthetic)4 Coding and Decoding Process The process of writing and reading, or coding and decoding a written text like Kojiki, was a rather complicated one and can be expressed by the following scheme:

2  Synthetic languages are those in which syntactic relations are expressed by means of inflections or by agglutination. Latin is an example of a typical inflected language; Japanese is an example of an agglutinative language. 3  Analytic languages are those that instead of inflections, express syntactic relations by means of autonomous lexical elements. Typical examples are Vietnamese and Classical Chinese. 4  OT1: initial oral text to be written in the Japanese language. OT2: fijinal oral text produced by “reading” in the Japanese language. HK: Hentai kanbun (simil-Chinese).

symmetry and asymmetry

173

OT1 (LS1)  >  yōjihō  >  OT2 (LS2) (wk) (kd) ↓  ↑  WT (LSwt)5 That is from an oral text OT1 with its proper language structure (LS1), by means of the strategies of use of characters ( yōjihō) a written text is produced in waka kanbun (wk), that is a hybrid language. From the written text, by means of kundoku strategies (kd)6 and passing through the strategies of use of characters the fijinal oral text (OT2), with its proper language structure (LS2), is produced. In the process of writing, many functional particles originally present in the Japanese oral text were lost in the written text and later reintegrated by the reader when reading. This happened because while a synthetic language such as Japanese needs to express syntactic relations in detail, Chinese and other analytic forms of language can often do without. Let’s see a simple example (taken from Kojiki): 1. OT1: Kono toki hashi sono kawa yori nagare kudariki. Translation: “At that time, chopsticks came flowing down that river”; 2. WT (Kojiki): 此時箸從其河流下; 3. OT2: same as OT1. In the written text of Kojiki, in the double verbal structure of the oral text, nagare kudariki “came flowing down”, -re of nagare, which is an inflected form of the verb nagaru “flow”, and -riki of kudariki, which is the past form of the verb kudaru “come down”, are not graphically represented. The reader must reintegrate them when reading. It is as if the English corresponding sentence “At that time, chopsticks came flowing down that river” was written “At that time, chopsticks come flow down that river” and the reader reads it the correct way.

5  LS = Language Structure; WT = Written Text; Wk = waka kanbun: written codifijication; Kd = kundoku: oral decodifijication; yōjihō (用字法): strategies for the use of characters. 6  Kundoku is the process of reading a Chinese text in Japanese by means of glosses (called kunten 訓点) inserted in the text between the vertical lines in order to help reading. The process of reading Chinese in Japanese was performed with two parallel strategies: recollocations of the word order according to Japanese language (SOV) and integration of the needed functional parts. This was carried out with the help of kunten.

174

aldo tollini

This means that, very often, the efffort of imitating the Chinese model produced the efffect of leaving parts of the oral language OT1 graphically unrepresented in writing. Actually, there were some not minor diffferences in the language structure between the oral language and the written language. For example it was common in Kojiki to have a fluctuating sentence structure with SOV and SVO patterns.7 This means that the language structure was not fijixed. There was a great gap between the oral language (always SOV and with many inflections) and the written language (which could also be SVO and with few inflections) at various levels. Therefore, considering what was stated above, we may wonder if the written language actually represented the oral language, or vice versa, if the oral language represented written language. The Lexical Problem (wakun) As described synthetically above, one of the main strategies employed by the Japanese in order to write their language was the so-called wakun, that is giving a Chinese character/word the Japanese equivalent. In other words, they operated a translation giving to a Chinese character/word the Japanese equivalent or near equivalent. For example the Chinese character/word 山 which means “mountain” was used for the Japanese word yama that has the same meaning. However, some problems arose because the Chinese and Japanese lexicon did not coincide. The gap between the lexicon of diffferent languages is a well-known phenomenon that causes problems in translation. In the case of Japanese writing, the use of Chinese characters/words made this problem even more complicated. In the case dealt with here we can list the following discrepancies between the lexicon of Chinese and Japanese: 1. Japanese words with more than one equivalent in Chinese; ( J1:Cnx) 2. Japanese words without equivalents in Chinese; ( J1:C0) 3. Chinese words/characters with more than one equivalent in Japanese; (C1:Jx) 4. Chinese words/characters without equivalents in Japanese. (C1:J0)

7

 SVO: Subject-Verb-Object structure, SOV: Subject-Object-Verb structure.

symmetry and asymmetry

175

Japanese words with more than one equivalent in Chinese This is the case of a single Japanese word in front of multiple Chinese character/words. This caused multiple writing since multiple Chinese characters/words corresponded to a single Japanese word. In Kojiki we fijind many such examples. Let’s see a few: A Ahu (“meet”) written with three diffferent characters 1. 会; 2. 逢; 3. 遇. B. Nari (“be, become”) written with nine diffferent characters 1. 作; 2. 爲; 3. 成; 4. 生; 5. 變; 6. 化; 7. 臨; 8. 至; 9.結. C. Naku (“weep, cry”) written with three diffferent characters 1. 泣; 2. 哭; 3. 鳴. D. Omohu (“think”) written with eight diffferent characters 1. 欲, 2. 惟, 3. 以爲, 4. 爲, 5. 念, 6. 思惟, 7. 思, 8. 懷. E. Koko (“here”) written in six variations 1. 斯, 2. 是, 3. 焉, 4. 爾, 5. 此地, 6. 此間. The above Chinese characters/words in the fijive groups have all, respectively, the meanings of: A. “meet”, B. “be, become”, C. “weep, cry”, D. “think”, E. “here” though with diffferent nuances. Japanese had a single word that did not distinguish among the diffferent nuances of Chinese. In Kojiki sometimes the Chinese nuance was maintained for the Japanese word by means of a conscious selection of the character based on the Chinese nuance and the meaning in the text (the distinction was not at the phonetic level, but only at the graphic and semantic one), but in many other cases it was not. In general, in Japanese ancient literature we can observe a rather free use of Chinese characters in order to represent Japanese words. As a matter of fact, they used Chinese character/words for autochthonous words on a rather free basis in the case of multiple choice. As a consequence, the same word could be written in many diffferent ways. Chinese words with more than one equivalent in Japanese This is the opposite case: Japanese had multiple words for the same Chinese character/word. For example, in Kojiki to the character 生 (“be born, live”) at least six autochthonous words were assigned: 1. Umu, 2. Sei, 3. Naru, 4. Aru, 5. Ohu, 6. Iku whose meanings span from “live” to “be born” or even “to be” or “to become”. The rather free rendering of Chinese characters/words in the autochthonous language is one of the main causes that

176

aldo tollini

makes reading ancient Japanese texts difffijicult and in same cases unavoidably approximate. Notwithstanding the enormous work of philologists in this fijield we can still have reasonable doubts about the real reading in those times. It seems to us that when writing, the ancient Japanese were much more concerned with transmitting a message by means of Chinese characters than with faithfully representing their own language, except in special cases. After all, a text written by Japanese in Chinese or in an approximate Chinese language had a much larger reading public in East Asia than a text written to faithfully represent the autochthonous language readable by a very restricted public in Japan. This is one of the main reasons why the Japanese stuck stubbornly to the Chinese written model. After all, written Chinese was the lingua franca of East Asia at those times, the vehicle of culture and civilization, playing a role similar to that of Latin in Europe. Japanese words without equivalents in Chinese This is the case, for example, of proper Japanese names ( geographical names, personal names, names of the gods, and the like), that is, words only present in Japan. Normally, in such a case, a phonographic strategy was preferred, whether shakuon or shakukun. Examples found in Kojiki: 1. 久羅下那州多陀用弊流 → ku-ra-ge-na-su-ta-da-yo-he-ru), (“float like a jellyfijish”). Here each character is used phonographically using its sino-japanese reading (on), that is the shakuon strategy; 久 for the syllable ku, 羅 for ra, 下 for ge, and so on. 2. 宇摩志阿斯訶備比古遲 → u-ma-shi-a-shi-ka-bi-hi-ko-ji (name of a god). Here also each character is used phonographically using its sinojapanese reading (on), that is the shakuon strategy; 3. 天之御中主→ ame-no-mi-naka-nushi (name of a god). Here each character is used phonographically using its Japanese reading (kun), that is the shakukun strategy. 天 for the word ame, 之 for no 御 for mi, and so on. 4. 鳥髮→tori-kami ( geographical name). Here each character is used phonographically using its Japanese reading (kun), that is the shakukun strategy. 5. 妹阿夜 → imo-a-ya (name of a god). Here we have a mixed writing system: imo is a logograph, and a-ya is a phonograph (shakuon). The mixed writing system was very common in ancient texts.

symmetry and asymmetry

177

Apart from the above strategy, there was another strategy for writing Japanese words not present in Chinese: that of creating new characters. These characters are called kokuji 国字, or “national characters”, that is characters “made in Japan” and not imported from China. Presently, there are quite a few of them in Japanese language, mostly for names of vegetables or animals typical of Japan. However, some are also used for rather common words, since some Chinese characters underwent semantic shifts and consequently even some common words needed new characters that were then invented in Japan. Normally these characters have only the Japanese reading (with a few exeptions) and were created only in ancient Japan but continue to be used nowadays. A few examples (not present in Kojiki) present in modern Japanese: 働 = “work” (read: hataraku); 峠 = “mountain pass” (read: tōge); 榊 = “a kind of camellia” (read: sakaki); 畠 = “fijield” (read: hatake); 凩 = “wintry wind” (read: kogarashi); 鰯 = “sardine” (read: iwashi); 辻 = “crossroads” (read: tsuji). Chinese words/characters without equivalents in Japanese In this case, the Japanese had to fijind a Japanese equivalent or introduce the new word from Chinese without change. In the second case they just read the Chinese word according to sino-japanese reading. In Kojiki we fijind the following sequences of characters in Chinese: 1. 天皇 read: sumera-mikoto “emperor” (“heavenly emperor”). A Chinese word to which the Japanese gave an autochthonous equivalent. In ancient Japan there was not ( yet) an “emperor”, nor its concept. Therefore, the Japanese used the word sumera-mikoto. Sumera derives from the verb suberu or sumeru which means “rule over a unifijied nation” and mikoto is a word used as a honorifijic for gods or persons in a very high position. 2. 典教 read: tenkyō “Buddhist teaching in the sūtras”. A Chinese word introduced in Japan and read with the sino-japanese reading. It is a direct loan from Chinese. At that time there was no possible word for the newly imported religion from China, so the Chinese word was used.

178

aldo tollini

3. 乾坤 read: kenkon “universe”. A Chinese word introduced in Japan and read with the sino-japanese reading. The Japanese had no word for “universe”, so they used the Chinese word kenkon, or otherwise, a calque of a Chinese word meaning “heaven and earth” (ametsuchi 天地). In Kojiki we fijind both words. The problem of writing was not limited to the lexicon. For example, also functional words had to be represented. While Chinese uses few functional words, Japanese is a language with a detailed representation of word function and word relations. Representing these words with kanji was a difffijicult task, and often they were not represented and left to the integration capabilities of the readers. The Japanese used phonographic strategies, or alternatively, in order to enhance the Chinese flavour of the text, they used strategies like that for the Japanese topic marker は (ha, later read wa). In order to express this function, the Japanese took the Chinese character 者 (read sha, mono), which had a similar function in Chinese, and gave it the reading (and function) of ha. The character 者 in Japan was used as: 1. Topic marker ha; 2. Ba as nominalizer of a verb (rentaikei + ba); 3. Particle of time, or cause (izenkei + ba). From what is stated above, we can now give a general scheme of the main possible uses of Chinese characters in ancient Japan (all of which are present in Kojiki): 1. logographic use: ex. 神 for “god”, 天 for “sky”; 2. phonetic use 1 (shakuon): ex. 伊邪那岐 (伊: i, 邪:za, 那:na, 岐:gi, Izanagi) (name of a god); 3. phonetic use 2 (shakukun): ex. 鳥髪 (鳥:tori, 髪:kami, Torikami) (geographic name); 4. functional use: ex. 爾 (koko, “here”), 然 (shika, “like this”), 如此 (kaku, “like”); 5. Special uses: ex. 春日read Kasuga, 大長谷 read oho-hatsuse. The so-called “special uses” refer to the uncommon use of Chinese characters, such as the attribution of Japanese words to sequences of Chinese

symmetry and asymmetry

179

characters based, for instance, on poetic allusion. These strategies form a separate fijield of research. I limited this study to lexical problems, but, of course, there are also grammatical and syntactical problems that the ancient Japanese had to solve when writing Kojiki, but they are a matter for future research. References in Western languages Bentley, John R. 2001. A Reference Grammar of Early Old Japanese Prose. Leiden-BostonKoeln: Brill. Gelb, Ignace J. 1952. A study of writing. Chicago: University of Chicago Press, Habein Yaeko Sato 1984. The History of the Japanese Written Language. Tokyo: University of Tokyo Press. Miller, Roy Andrew 1980. The Japanese Language. Tokyo: Charles E. Tuttle Co. Olson, David R. 1993. How writing represents speech. Language and Communication, 13(1):1–17. Philippi, Donald L. (ed) 1969. Kojiki. Tokyo/Princeton: University of Tokyo Press & Princeton University Press. Reischauer, Edwin O. & J.K. Fairbank 1960. East Asia—The Great Tradition. Boston: Houghton Miffflin Co. Seeley, Christopher 1991. A History of Writing in Japan: Leiden: Brill. Shibatani Masayoshi 1991. The Languages of Japan. Cambridge: Cambridge University Press. Villani, Paolo (ed) 2006. Kojiki: un racconto di antichi eventi. Venezia: Marsilio. Vovin, Alexander 2005. A descriptive and comparative grammar of western old Japanese. Folkestone: Global Oriental. —— 2003. Reference grammar of classical Japanese prose, London, New York: Routledge Curzon. Wehmeyer, Ann (ed) 1997. Kojiki-den. Ithaca: Cornell University.

WRITING SEMITIC WITH CUNEIFORM SCRIPT. THE INTERACTION OF SUMERIAN AND AKKADIAN ORTHOGRAPHY IN THE SECOND HALF OF THE THIRD MILLENNIUM BC* Theo J.H. Krispijn Introduction The oldest written examples of cuneiform script were found by German archaeologists on clay tablets from the Eanna temple complex in the ancient city of Uruk during excavations in the 1930s. These tablets exhibit some development in their use of script but, since the stratigraphy of this area is extremely complicated and confusing, scholars have had serious problems in dating the documents that are known. But Sürenhagen (1999), after examining the original fijield notes of the excavators, has recently proposed a more balanced relative chronology for them. – Uruk V: numerical tablets ± 3400 BC – Uruk IVc–b: numerical tablets with logograms ± 3300 BC – Uruk IVa: a more developed script, incorporating logograms and phonetic signs (phonograms), used for administrative and lexical texts ± 3200 BC – Uruk III(c): an even more developed script, following the canonization of lexical series. ± 3000 BC. Texts of this type have also been found at Jamdat Nasr, Tell Uqair, and possibly at Larsa. Versions of the canonized lexical series of Uruk III (see Englund & Nissen 1993) continued to be reproduced till the Early Old Babylonian period (± 1800 BC), so it is possible to read the earlier versions with the help of the later versions, where the script is more easily readable. Phonograms clarifying logograms that we are thus able to recognise can determine the language of these documents. Both lexical texts and administrative documents of the Uruk III period for the most part appear to have been

* I am much indebted to Mervyn E. Richardson for improving the English of this paper and for some valuable suggestions concerning the paper itself.

182

theo j.h. krispijn

composed in Sumerian, and any use of logograms for phonograms can be explained as a Sumerian rebus:1 – Logogram E2 (/hai/) “house” for the phonogram /ha/ – Logogram GI “reed” for the phonogram /gi/ “to return, to hand over” later written (šu) gi4 – Logogram ZU/SU “flesh” for the phonogram /su/ While we can be reasonably certain that cuneiform was used for Sumerian texts from the Uruk III period onwards, as is especially clear from literary texts from the Fara period (± 2600 BC) which have been handed down in later traditions, it is more difffijicult to establish the language of the texts from Uruk IVa. Only a few phonograms were used in this earlier stage of the script, leading some scholars to think that the language of these texts is also Sumerian while others have serious doubts about that idea.2 The question of the presence of Semitic elements in the cuneiform texts of the Uruk IV/III and subsequent Early Dynastic (ED) I/II Period (3000–2600 BC) is intriguing. Fifty words (14 from the Uruk period and 36 from the ED) have been proposed as Semitic loanwords and names in scholarly literature. But Sommerfeld (2006) has critically discussed all these assumed Semitic elements and concluded that the only loanword that can undoubtedly be regarded as a Semitic loanword is mana “halfkilogram”. Such diverging opinions among Assyriologists make it clear that we are in the middle of a lively discussion about the earliest evidence for Semitic writing. The ED III (2600–2400 BC) period gives a clearer picture concerning Semitic elements. Several Semitic loan words and Semitic (Early Akkadian) names occur in texts from Fara (Šuruppak), Abū Ṣālābīkh (Krebernik 1998, 260–270) and Telloh (Ĝirsu) (Bauer 1998, 437).3 The influence

1

 The conventions of the main text are: Bold face for Sumerian and italics for Akkadian phonograms, morphemes and lexemes. Capitals indicate a ‘sign name’, i.e., a transliteration of a sign without explicit phonemes that are arbitrarily chosen from all possible transliterations of a sign. 2  Englund (1993, 80–81) assumes that the texts of the Uruk IVa render a language different from Sumerian, but Krebernik (1994) thinks that the language of this period was already Sumerian. The sign AMA could support the Sumerian hypothesis. AMA is a picture of house inscribed with a star and can be interpreted as a depiction of a storageroom (ama5). The inscribed star AN must be read am6 as phonetic indicator for the logogram ama5, that served as phonogram for ama “mother”. 3  The important article by W. Sommerfeld (2010) came too late to use it properly for this article.

writing semitic with cuneiform script

183

of Semitic on Sumerian is obviously increasing in this last stage of the ED period but it is not always easy detectable. M. Civil (2007) has made a long list of Sumerian lexemes with two syllables of which a high percentage most probably originate in Semitic. A number of these Semitic loans are already attested as lexemes in the ED period. This is further confijirmation of our uncertainty about the influence of Semitic, or more precisely (Early) Akkadian, before the Sargonic period (2350 BC), referred to earlier. But we fijind ourselves on more solid ground with larger groups of cuneiform texts where Akkadian (Eastern Semitic) is written together with Sumerian for the fijirst time. The large number of available Sargonic Akkadian and EblaAkkadian (EbAkk) texts make them particularly suitable for our purpose of studying the interconnections in the use of cuneiform script for writing both Sumerian and Akkadian. Since, at the end of the third millennium, Sumerian is considered to be a living language with a spoken vernacular and a written form, and since, at the beginning of the second millennium, it was spoken in the Old-Babylonian scribal schools (Michalowski 2006; Woods 2006), we can assume that the distinction of the phonemic Systems of Sumerian and Akkadian was still observable. I have chosen Sargonic Akkadian (SargAkk) as the starting point for this presentation of my research on the interaction or non-interaction of Sumerian and Akkadian orthographies. I follow this with an examination of the slightly earlier Pre-Sargonic Sumerian material. In the third section the orthography of Ebla-Akkadian will be discussed, and in the fourth the phononetic writing, i.e., with phonograms, of Sumerian in Ebla. The various tabulations of orthography illustrate how Akkadian and Sumerian phonemes were rendered in CV and VC signs.4 Selected observations from the diffferent sections concerning the spelling of Akkadian and Sumerian are presented in the fijinal synthesis to demonstrate the level of coherency in the spelling of both languages.

4

 The conventions of the tables are: – Capitals (column 1, 3, 5, 7) are used for ‘sign names’. – Homophones are numbered in subscript according to the assyriological conventions established in Borger 2003. Phonemic reconstructions are marked with double slash. – There are 2×4 columns for the combinations Consonant +A; Consonant +E; Consonant +I; Consonant +U and reverse. The tables for EbAkk and Sumerian have 3×3 and 1×3 columns. – The symbol H is used to indicate a fijinal consonant or (strong) hiatus following or preceding a vowel. When H follows a vowel this vowel is short.

184

theo j.h. krispijn The Syllabary of Sargonic Akkadian

Sargonic Akkadian (SargAkk) Akkadian was fijirst written on a large scale when King Sargon (± 2350 BC) made the language of his capital, Akkade, an alternative offfijicial language to Sumerian in southern Mesopotamia. It was mainly introduced for writing administrative texts and letters, but some royal inscriptions and incantations were also composed in a literary form of Akkadian. In her recent grammar on SargAkk Rebecca Hasselbach pays considerable attention to the syllabary and phonology (Hasselbach 2005: 32–146). Because the general principles of the syllabary are consistent SargAkk is treated here as a unity. However, there are some local diffferences in orthography and dialect which occur, especially between texts from the Diyala region, Southern Babylonia, and Northern Babylonia. Sometimes sign pairs, i.e., phonograms, which might be used interchangeably, do not reflect phonemic opposition but diffferent local orthographical conventions (Hasselbach 2005: 39–73).5 In the tables below I usually follow Hasselbach’s reconstruction of Old-Akkadian phonology but with some minor modifijications. Orthographic conventions The orthography of SargAkk is a mixed system, in which phonograms and logograms are combined. For the greater part it can be seen as a development of Pre-Sargonic Northern Akkadian orthography (Hasselbach 2005: 30–31). In a word written phonetically usually all the phonemes are explicitly indicated except for features such as consonant doubling and vowel length. Closed syllables are written either with two signs (CV+VC) or one sign (CVC), but not all the CVC-signs are discussed here. Signs indicating a short vowel in a closed syllable form a special group (see below). The phonograms: consonants Almost never a distinction is made between voiced, voiceless and emphatic consonants (e.g., /d/ :: /t/ :: /ṭ/) in the syllabary. The reconstruction of voiced and voiceless consonants is based on phonemic oppositions

5  Local orthography: DU = du, ṭu3 (North Babylonian) :: TU = du, tu, ṭu2 (SargAkk dental+u in general). Presargonic orthography: NE = de3, te4, or ti9 (see notes to dentals etc.); SA :: DI = sa2 (see also notes to palato-alveolars).

writing semitic with cuneiform script

185

in Proto-Semitic which are clearly marked by correspondingly diffferent signs in the orthography after the Ur III period. The reconstruction of voiced, unvoiced and emphatic stops6 is similarly based on Proto-Semitic oppositions reflected in writing after the Ur III period. The precise articulation of emphatic stops in SargAkk is not entirely clear. Diverse articulations are attested within Afro-Asiatic languages, to which Semitic languages belong. That they were postglottals (ejectives) in Old-Assyrian, a later dialect of Akkadian, and not pharyngeals as in Arabic has been demonstrated by N.J.C. Kouwenberg (2003).7 A postglottal pronunciation can also be supposed for SargAkk but cannot be proved. It is assumed that signs of the Š-series are also used for interdental /ṯ/ combined with a vowel, e.g., ŠA = /ṯa/, an assumption made plausible because the series is used for /ṯ/ as well as /ḏ/ in Ebla Akkadian. Directly after the Sargonic period /ṯ/ is spelled both with signs from the Š-series and the S-series, especially in the environment of /i/ and /u/ in Northern Babylonia (Hasselbach 2005: 143).8 The Z-series renders the phonemes z, s, ṣ and ḏ. In SargAkk, in contrast with Ebla Akkadian there is no demonstrable distinction between ḏ :: z (see below, the note to SU4), for both phonemes are written with the Z-series. ṣ represents a coalescence of three Proto-Semitic emphatic phonemes traditionally transcribed as /ẓ/, /ḍ/ and /ṣ/. It is widely accepted that z, s and ṣ were all pronounced as afffricates, /dz/, /ts/, /ts’/ (Hasselbach 2005: 95–97+199–201). Reconstructing the pronunciation of the S-series is difffijicult and three possibilities are offfered: – as /š/ it accords with Proto-Semitic /š/ and the later Old-Babylonian pronunciation – as /ś/, a lateral fricative, it explains the later development /št/ = /śt/ > lt 6  The following emphatic stops and fijicatives are reconstructed in Proto-Semitic: dental stop: ṭ; interdental fricative: ẓ; lateral stop: ḍ; velar stop: ḳ or q. This last stop is realized as uvular voiceless stop in most Semitic languages. 7  See also Steiner (1982) and Faber (1985: 101–107+7). 8  Outside Semitic (especially in Hittite) the Š-series of the cuneiform script is often used for etymological *s (Kloekhorst 2008, 89–94). In Ugaritic alphabetic script ṯ represents the Akkadian syllabic cuneiform Š-series This Š-series is mostly rendered by ṯ in Akkadian texts written in Ugaritic alphabetic script, but also with š (Segert 1988, 191). Syllabic Hurrian distinguishes V-ŠV :: VŠ-ŠV that is reflected by -ḏ- :: -ṯ- in alphabetic Hurrian (Laroche 1968, 528).

186

theo j.h. krispijn

– as /s/ it accords with the writing ZU for the assimilated consonantal cluster in /t+šu/ > /tsu/ I favour /š/, since most older Semitic languages have /š/ corresponding to the Akkadian S-series, e.g., Akkadian u-sa-lim = yušallim (cf. Hasselbach 2005: 284) < šlm (Ugaritic, Hebrew, Phoenician, and Aramaic) but slm (šlm Arabic, South-Arabic, and Ethiopic, although Hasselbach prefers /s/ (Hasselbach 2005: 96–97).9 The following ‘sign pairs’ reflect phonemic oppositions with or without a strong hiatus, such as gutturals, here indicated with H. Or, they reflect oppositions with a soft hiatus such as /y/, here indicated with y: – phonograms with or without H: MA = ma :: MA2 = maH; AM = am :: AM3 = /Ham/; LA = /la/ :: LA2 = laH (see notes to dentals etc.). – phonograms with or without y: IM = im :: IM4 = /yim/; IL = il :: IL2 = /yil/, IR /ir/, /er/ :: IR3 = /yir/, /yer/. – phonograms with or without /H/-/y/: EN = /yin/, :: IN /Hin/, /in/. It is interesting that signs like EN are polyvalent, indicating initial /y/ or /H/. The phonograms: vowels The phonemes /e/ is imprecisely distinguished from /i/. The phoneme /e/ may result from the contraction of /ay/ or /ay( y)i/ (Hasselbach 2005: 119), or from the lowering of /i/ in the environment of /’/ or /r/, or from the raising of /a/ in the environment of /’/, /r/, /ḥ/ or ‘ ( yušōṣē (u-su-ze2) “he gave out” Hasselbach 2005, 224. 12  The numbers in the colums refer to: 1. signs C+A or A+C. 2. phonemic reconstruction of the signs C+A or A+C. 3. signs C+E or E+C. 4. phonemic reconstruction of the signs C+E or E+C. 5. signs C+I or I+C. 6. phonetic. reconstruction of the signs C+I or I+C. 7. signs C+U or U+C. 8. phonetic. reconstruction of the signs C+U or U+C. Phonograms marked with G are attested in Lagaš-Ĝirsu in the Sargonic period. Special attention is to be paid to Lagaš-Ĝirsu, since it is an almost continuous source of Sumerian documents from ca. 2500 to 1800 BC. 13  Gelb (1961, 58) mentiones some PN’s with BAD beside those with ba‘lum it-BAD-um = jitba‘um and ša-at-BAD-diĝir = ṯatpi-‘il (Diyala Region).

188

theo j.h. krispijn

Dentals, interdentals, laterals and rhotics 1 DA (G)

2 da, ta2, ṭa

3 TE

4 te

5

ad, at na an

EN (G)

ŠA (G) /ṯa/ ŠE3 AŠ b (G) /aṯ/, /Haṯ/

/yin/, /Hin/ /ṯe/

7

8

TI (G)

di3, ti, TU (G) te9 DU (G)

du, tu, ṭu2 du, ṭu3

ID (G) NI (G) IN (G)

id, it ni in

UD (G) NU (G) UN (G)

ud, ut nu

/ṯi/ /iṯ/, /yiṯ/ /iṯ/, /yiṯ/ li2

ŠU (G) UŠ

/ṯu/ /uṯ/

LU (G)

lu

NE (G) a de3, te4, ti9 AD (G) NA (G) AN (G)

6

EN ŠI (G) IŠ11 (G) EŠ3

LA (G) LA2 c AL (G)

la /laH/ (?) al

LI (G)

le, li

NI (G)

EL

el

IL (G) IL2 (G)

RA (G) AR (G)

ra ar

RI (G)

re, ri

il /ỉl/, /yil/ URU (G) ri2 IR (G) ir, er IR3 (G) /yir/, /yer/ IR11 e (G) /yir/

UL d (G) RU (G) UR (G) UR4

ru ur ur

Notes: a The combination of ṭ+e/i/u, which occurs in Ebla, is not attested due to the limitations of the Sargonic text corpus. NE is not used for ne but only for de3, te4 and bi2, pi2. The use of NE = de3, te4, or ti9 is restricted to traditional spellings (Hasselbach 2005: 54–57). b AŠ for /‘aṯ / especially in ‘aṯtar “Ištar”. c The attestations of LA2 are too few to draw sound conclusions on the oppositions LA :: LA2. By analogy with the Ebla material it may correspond to /la/ :: /laH/. d UL in si-tu-ul-ni-se11 PN ITT 1 1062. e IR11 = /yir/ in ir11-ti-ab “he has replaced” ITT 1 1103.

writing semitic with cuneiform script

189

Palatals and alveolars 1

2

3

4

5

6

ZA (G)

za, sa3, ZE2 (G) ze2, se2, ZI (G) zi, si2, ṣa, (/ḏa/) ṣe2, (/ḏe/) ṣi2, (/ḏi/)

AZ (G)

az, as, aṣ, (/aḏ/)

SA (G) b sa = /ša/, SIG (G) se11= /še/ (/ṯa/) DI (G) b sa2 ŠE3 (G) /še/ AŠ2 (G)

as2 = /aš/

7

8

ZU (G)

zu, su2, ṣu2, (/ḏu/) KA a (G) zu2 IZ (G) iz, is, iṣ, UZ uz, us, iḏ uṣ, uḏ EŠ c (G) /uz/ SI (G) si = /ši/ SU (G) su = /šu/ SU4 (G) su4 = /šū/ IŠ (G) is2 = /iš/ UŠ (G) us2 = /uš/

Notes: a KA = zu2 in zu2-ba-lum = supālum “juniper tree” ITT II/2 4587 rev. b There is no phonemic contrast between SA2 and SA. SA2 is only used in Pre-Sargonic Akkadian and in Sargonic Ĝirsu and Royal inscriptions. In later Sargonic texts SA renders not only /ša/, but also /ṯa/. c EŠ = /udz/ is limited to forms of the verb aḫāzum < ( ̉)aḫāḏum “to take”, e.g., li-ḫu-EŠ = līḫuz RTC 77 rev., but it is very unlikely that it reflects /ḏ/ since that phoneme coalesced with Z = /d z/ before the Sargonic period.

Velars, uvulars, gutturals, and pharyngals 1

2

3

GA (G)

ga, ka, qa2

GI (G)

AG ḪA (G) UḪ (G)

ag, ak, aq ḫa, /ǵa/ aḫ

A (G) A2 (G) E2 (G)

/’a/ /ha/ ’a3 = /ḥa/, (/ha/) /ha/

E3 c (G)

4

5

6

7

8

ke2, qe2 KI (G) ki, qi2

E b (G) e, /’i/

GU (G) a gu, (ku8), qu3 GI4 ki KU (G) a ku, qu2 IG (G) ig, ik, iq UG uq ḪI (G) ḫi ḪU (G) ḫu UḪ iḫ UḪ uḫ UḪ2 uḫ2 NI (G) i3 = /’i/ U2 (G) /’u/, /yu/ U3 (G) u3 =/’u/ E (G) /Hi/

Notes: a The opposition GU :: KU renders mainly /qu/ :: /ku/ (Hasselbach 2005: 60–62). b E = /He/ i.e any guttural +e. Beside that usage it indicates simply vowel quality /e/ (Hasselbach 2005: 75). Cf. also e-la = ỉla PN ITT 1 1460. c E3 = e3-dib-me-er = hātip-mer PN RTC 127 rev. VI.

190

theo j.h. krispijn

Semi-vowels and vowels 1 NI+A (G) a A (G) PI (G) b (U+A)

2

3

4

5

6

7

8

i3-a = /ya/ /ya/ wa (/wa/)

E E3

/ye/ /ye/

I (G)

/yi/

U (G)

/yu/

Notes: a NI+A = /ya/ li-pu-uš-NI+A-um = līpuš-ya’ûm PN DC II pl LII. b There are no attestations for PI = /we/, /wi/, /wu/ in Sargonic Akkadian.

The Syllabary of the Pre-Sargonic Sumerian in Lagaš Pre-Sargonic Sumerian (PreSargSum) in Lagaš Sumerian is an isolated language with no afffijiliation to a known language family. So the phonograms cannot be reconstructed on the basis of comparison with cognate languages or (external) etymology. Fortunately we have other sources at our disposal such as Sumerian logograms used as phonograms in Akkadian texts, Sumerian loanwords in Akkadian, phonetic glosses to Sumerian logograms in lexical and literary texts, and even the names of signs. The most important source for the pronunciation of Sumerian in the Old-Babylonian period is the syllabary Proto-Ea, which contains single logograms with phonetic glosses and sometimes an Akkadian translation (Civil 1979). On the one hand such sources make an internal reconstruction of Sumerian possible, but on the other hand the writing system complicates the reconstruction by hiding compounds or loanwords behind single signs, or by using diffferent signs for the same Sumerian words in an artifijicial way.14 For the reconstruction of a Sumerian phonograms slightly earlier than the SargAkk system we have at our disposal the PreSargSum texts from Lagaš-Ĝirsu. The language of those texts is relatively well known thanks to the large amount of tablets found at Telloh (Krispijn 2005). Among the texts found were votive and building inscriptions of the city rulers and the archives of the consorts of the last three rulers. Most of these texts date from the end of the Pre-Sargonic period, the last fijifteen years of Lagaš 14  An example of a compound word written with one sign is: ŠAGAN “a bottle for oil” < šag4 “belly” + gan “thick, pregnant”. An example of an artifijicially diffferentiated sign for one word is: UGU “top of the head” :: AGA “crown (head dress)”.

writing semitic with cuneiform script

191

as an independent state, so the orthography of the Sumerian is directly comparable with that of the Akkadian of the Sargonic period. Traces of a special emesal-like dialect of Sumerian as spoken in Lagaš have sometimes coloured the offfijicial Sumerian style (Krispijn 2005, 154–156, 161). Orthographic conventions PreSargSum is written partly with phonograms and partly with logograms. Phonograms are mostly used for the verbal prefijixes and sufffijixes and nominal morphemes (such as postpositions and pronouns), non-Sumerian names and loanwords. CV +VC od CVC signs may be used to render closed syllables, but usually the last phoneme of a closed syllable is not written. Sumerian has special orthographic conventions for the end of a word, where case postpositions and pronominal sufffijixes follow a word base. The fijinal consonant of a word base, which is mostly written as a logogram, is repeated in some phonograms, a spelling feature which is labeled here as ‘resumptive’, e.g., BARAG.GA = barag+a. A spelling when the last consonant is not repeated is labeled here as ‘broken’, e.g., TUŠ.A = tuš+a. The phonograms: consonants That the Sumerian phonograms difffers considerably from the SargAkk system will be obvious, but details need to be listed. Instead of the three variations of stops of SargAkk (voiced, voiceless, ‘emphatic’), Sumerian has only two and although their character cannot be determined with certainty, they are traditionally transliterated as voiced, e.g., da :: voiceless, e.g., ta. Information for the character of the consonants at the end of a word can be deduced from resumptive and broken spellings (see above). Resumptive spellings can be categorised as follows: – +a (1) fijinal consonant + locative postposition or defijinite morpheme (comparative-superlative) / past participle marker a: šag4.ga = šag4+a “in the heart”; zid.da = zid+a “the (more/most) favourable”; dab5.ba = dab5+a “taken”. – +a (2): fijinal consonant + /a/ of the genitive postposition a(k) or the possessive pronoun ane “his, her” and *anene “their”: šag4.ga.(k= šag4+a(k) “of the heart”; šag4.ga.ni = šag4+ane “his heart”. After some consonants, especially before other morphemes, this /a/ is not written, After vowels it is also deleted, having weakened to a šewa-like pronunciation indicated by ø.

192

theo j.h. krispijn

– +e (1) fijinal consonant + e of the directive or ergative postposition,15 the plural morpheme ene or the pronominal ending after the imperfective verbal base (marû): dam.me = dam+e “by the husband”; -ḫa.lam.me = ḫalam+e “he will destroy”. +/e/ (1) is found as broken spelling more often than +/a/ (1), especially with pronominal sufffijixes after the imperfective base (marû). – +e (2) fijinal consonant + e morpheme of the imperfective verbal base: dim2.me.de3 = dim2+e+d+e “(in order) to fashion”. This /e/ often assimilates to the vowel of the verbal base, again after weakening in quality to become šewa-like in quality. The pronominal ending after the perfective verbal base (ḫamṭu), incorporating the absolutive, i.e., an object of transitive verbs or a subject of intransitive verbs, most likely contains e (2). – +u < e (2) assimilated to /u/ of the base, in combination with the pronominal ending e (1): -sum.mu = /sumu/ < sum+e (2) +e (1). Note that /u/, and to a lesser degree /i/, < /e/ (2) and assimilated to the /u/ or /i/ of the base, is not written: šum.de3 = /šumude/ < šum+e+d+e “(in order) to slaughter” dim2.de3 /dimide/ beside dim2.me.de3 < dim2+e+d+e “to fashion”. A possible exception is gur.ra.da.am6 (see notes to dentals), where /ru/, assimilating /r+e/ (2) after gur, is written with RA. Phonograms at the end of words are tabulated below. The following terms are used for the rate of the occurences: full = 100%; high = between 100– 50%; equal = 50%; low = between 50–0%; hapax = 1x; zero = 0%. The resumptive spelling is attested with: – C+a (1) stops: BA (high), DA (full), DU = řa2 (full); ZA (no attestations), GA (full); nasals: MA (full), NA (full) ĜA2 = ĝe26 (full); rhotics: RA (high); fricatives: -Š.SA (low), Š.ŠA4 (low), ḪA (high); laterals: LA (high). – C+e (1) stops: BI = be2 (equal), NE = de3 (high), DU = ře6 (high), (ZEx), GE (high); fricatives: ḪI = ḫe (equal); nasals: ME (equal), NI = ne2 (high), ĜA2 = ĝe26 (full); rhotics: URU = re2 (high); laterals: LI = le (high). – C+u (1) stops: GU (hapax) nasals: MU (full); laterals: LUL = lu5 (full).

15  The anaphoric pronoun e “this” , also with e (1), is not attested in Presargonic Sumerian.

writing semitic with cuneiform script

193

The broken spelling is attested with: – C+a (1) stops: -P.A (low) rhotics: -R.A (low); fricatives: -Š.A (high), -Ḫ.A (low); laterals: L.A (low). – C+e (1) is attested with stops: -B.E (equal), -Ř.E (low), -Z.E (hapax), nasalized consonants: -M.E (equal), -N.E (low); rhotics: -R.E (low); fricatives: -Š.E (full), -Ḫ.E (equal); laterals: -L.E (low). – C+u (1) is not attested (zero). For the greater part the following morphemes are not rendered in writing: – The genitive postposition ak (2). – The morpheme e (2) (imperfective verbal base). – The morpheme u (2), the assimilated e (2) morpheme of the imperfective verbal base. It is clear that stops, nasals, rhotics, and laterals have a high percentage of resumptive spellings especially with C+a (1) and to a lesser degree with C+e (1). Only fricatives have more broken spellings. As mentioned above, closed syllables are, in contrast with SargAkk, rarely written with the combination CV-VC phonograms and only a few VC phonograms are in use in PreSargSum. That is why we have only scarce information about the fijinal consonant of closed syllables. In Sargonic Sumerian and later there is a growing tendency to write the fijinal consonants. Some of the signs transliterated here with a voiceless stop have special grammatical functions: – The sign TA is used only for the Sumerian ablative postposition ta and the case element ta in the verbal prefijix chain. Therefore it is to be considered a type of logographic writing. – The sign KA = ka is used for the combination of the genitive ak + a ( genitive or locative) and in the spelling of the perfective base of the verb AK “to do”+ a (1): AK.KA = /ak+a/. – KID = ke4 is used for the combination of the genitive ak and the ergative or directive postposition e. It is also attested in the spelling of the imperfective base (marû) of the verb AK “to do”: ak+e (1) > /aka/ +the pronominal element 3S ē: AK.KE4 = /(a)kē /.

194

theo j.h. krispijn

More about the pronunciation of the p, t, and k can be said on the basis of texts from later times: – p: PI is used in Sumerian texts as an alternative writing for bi “that, its” and for /be/ in PI-lu5-da = belludû. In Akkadian texts PI is used for w-series. up2+a is written with broken spelling, which might indicate /p’/, with a possible allophone [pf] which must have been close to bilabial /w/. – t: In later periods (Gudea/early Ur III) the case element ta can often be written -ra- especially after ba-/im.ma-/mu.na/-ra-/-da- (See also Gragg 1973, 93–98). In close contact with /r/ the genitive or locative is not spelled RA but sometimes TA, e.g., sahar.ra/ta, sirar(a).ta, instead of sirar(a).ra.16 The pronunciation of TA was apparently close to /ra/. It needs further study to determine whether /t/ of TA coalesced with or was allophone of /ř/. – k: The early Semitic loanword /maška’en/ < /muškacin/ is written MAŠ. EN.GAG = MAŠ.GAG.EN. The last velar of GAG renders apparently the pharyngal /c/. Another interesting phenomenon is the pronunciation of the genitive morpheme ak before consonants. In PreSargSum the dative morpheme ra, which becomes /r/ after vowels, is seldom written after words ending in a vowel.17 By contrast, in words ending in a consonant the morpheme is always written. After the frequently occurring compound name Ninĝirsu(k) (= (n)in+ĝirsu+ak “Lord of Ĝirsu”) however, the dative morpheme is never left out. Although ak was never written at the end of a word in the Pre-Sargonic or subsequently, it was apparently pronounced as /’/ causing the full pronunciation of the dative morpheme ra. That makes it plausible to reconstruct the pronunciation of the genitive morpheme as /(a)k’/, or /(a)’/. In the light of the above observations a reconstruction of p, t, k as postglottalized stops: /p’/, /t’/, /k’/ seems more plausible than as aspirated stops ( Jagersma in Attinger forthcoming, p. 10).

16  saḫar.ta tuš “to sit in dust” Enki and Ninḫursaĝ 220; saḫar.da/ta šar2 “to mix with dust” GiAk 77, 95 (ra); Lugale 263, while Inanna’s Descend has locative saḫar.kur.ra.ka šar2. Sirar(a).ta.k Gudea A IV 5 which is the resumptive spelling of the GN Sirar(a)t+ak (genitive postposition). Cf. saḫar.ra/ta = sahar+a “in the sand”. 17  Words ending in -a: ama.(ra) “mother” Ent 79 IV 3; ama.(ra).gi4 “freedom” Ent 79 III 10 etc.; Ukg 4 XII 21 etc.; saĝa.(ra) “temple leader” Fö 120 IV 2; ugula.(ra) “chief” Fö 149 II 1; 178 IV 4. Words ending in -e/i: dNanše.(ra) a goddess Ent 26 12 etc. Words ending in -u: dumu. (ra) “son” Ent 79 IV 2; Fö 26 II 4; lu2.(ra) Ukg 6 II 6’.

writing semitic with cuneiform script

195

Some reconstructed phonemes are only attested in Sumerian and not in Akkadian, such as /ĝ/ and /ř/, others are comparable with Akkadian phonemes, but have a diffferent place of articulation. The following phonemes need further comment: – /ĝ/ was a nasal velar or nasal labio-velar, written with ĜA2-MI-MU.18 – /ř/, written with DU for /řa/ and /ře/, could have been a voiceless, aspirated, dental or alveolar afffricate [tsh] ( Jagersma 2000). This phoneme was indicated as /dr/ in earlier literature. – Of the two rhotic phonemes in Sumerian r (1), which is close to /ḫ/, represents an uvular /r/, and r (2) is a phoneme close to /l/. – /w/ is reconstructed on the basis of emesal equivalents.19 – The precise place of articulation of /š/ in the Sumerian Š-series is not known. Jagersma (Attinger forthcoming p. 10) suggests a palatal pronunciation (č?). /š+a/ and /š+e/, mostly written with broken spellings -Š.A and -Š.E, suggest a hiatus after the sibilant, possibly /h/. The fact that /š/ of the plural ending /eš/ (ŠE3 = eš2) is not written or replaced by a spelling with two vowels with hiatus -e.a20 supports this assumption, if one assumes that the sibilant-part of the phoneme was considered to be the closing consonant of one syllable, one often not written, and the h-part as the beginning of the next. In the light of the assumptions above the phoneme could be reconstructed as /čh/. The phonograms: vowels Sumerian very probably had long as well as short vowels. Long vowels influence the vowel of the verbal prefijixes in PreSargSum so that the prefijixes of some verbs have either e- or i-colouring (see for a complete inventory Krispijn 2005, 156 and 167). This phenomenon is usually called vowel

18

 Labio-velars can be reconstructed in: diĝir /diĝwir/ = ES dimmer/diner/dimbir “god” (cf. Veldhuis, J. ASJ 18 (1996), 230–231); eĝar /eĝwar/= ES abar “wall”; nimĝir /niĝwir/= ES libir “herold”. 19  e3 < *we/ye = ES mu.(da) “to go out” TLB 2 1 II 10 // TRS 5 No 5 5; EN < /ewen/ “enpriest” = ES umun Emesal Voc. II 7–9; eš/peš < *weš = ES (am3).mu.uš “three” Emesal Voc II 92 (cf also ab2:eš = ab-ma2-uš ED Animal List A 5 (MEE 3, p. 51 // p. 251 No 62 I 5 phon.) sa.eš5 = ES sa.muš Nabnitu XXXII i 2); lu2 < *wulu “man” (cf. lu2.ulu3 < *(wu)lu. wulu = ES mulu Emesal Voc. II 6; (ĝi6) u3.na < *wuna = ES (mi) mu.na “midnight” VS 10 179 1–2; unu3.d < *wunud = ES munu Emesal Voc. 13–14. See Diakonofff, 8948. 20  E.g., ba.ug5.ge = ba.ug5.ge.eš2 DP 138 II 2; ba.ug5.ge.a = ba.ug5.ge.eš2.a Fö 39 I 4; ba.de6.(*eš2) RTC 66 VI 4; mu.de6.(* eš2).a RTC 31 III 4; šu ba.ti.(*eš2).a Fö 88 I 4. Also emesal deletes sometimes /š/: ĝiš = ES mu “tree” Emesal Voc. I 105.

196

theo j.h. krispijn

harmony and is also operative within the prefijix chain, when certain prefijixes cause the preceding element to have e-colouring (see below, table on phonograms). Most verbal bases in /a/ have e-colouring, and most in /u/ have i-colouring. Sumerian has a strong tendency to contract diphthongs, so it is surprising that e (2) after -(C)A and -(C)U is often maintained, though never after -(C)E (see below, table on vowels). That is why it is attractive to reconstruct the ending *e as /ye/. Tables of the PreSargSum phonograms21 The tables below contain only CV and VC phonograms. The commentary on the individual signs cites special instances of signs relevant to this discussion. Labials C+a/ a+C

phonogr. C+e/ e+C phonogr. C+i/ i+C

phonogr. C+u/ u+C

BA PA a AB

ba – ab/

IB2

ib2

MI IM

mi im

MA ma AM am AN (a) am3

BI PI (a) b ME

be2 pe, pi me

phonogr.

NE (a) bi2 UB ŠE3 (a) MU (a) UM

ub up2 mu um

Notes: a PA is not used as phonogram, only logographically in pa.bil3.ga “grandfather, uncle (brother of the father) ” and pa “branch, stick”. The same Sumerian word written as pa4 ” branch of a canal” is used metaphorically to mean “grandfather”, i.e. a new branch on the family tree. b PI is only attested in the loanword PI-lu5-da “rituals” < belūt(um) “lordship” (Akkadian), where PI = pi or bi3. PI-lu5-da = /billuda/ = billudûm, a backloan from Sumerian, according to later Old-Babylonian sources.

21  The table has Sollberger 1961 as a startingpoint and adds data from more recent publications on PreSargSum from Lagaš. The abbreviations used in the notes are listed in Steible, H. 1982. Die altsumerischen Bau- und Weihinschriften I, II, Wiesbaden: 1982.

writing semitic with cuneiform script

197

Endings:

C+a (1)

C+a (2)

C+e (1)

C+e (2)

BA / -P.A c MA

BA / -B.∅

BI / -B.E d

BI

MA / -M.∅

ME / -M.E

ME /-M.∅

C+u (1)

C+u (2)

MU

-M.∅

c -P.A: til.lu.up2.a Fö No 177 I 3. d -B.E: na.ab.dab5.e Ean 62 IV 2 6’; nin-nam.mu.šub.e (PN) STH 23 9.

2.4.2 Dentals, interdentals, laterals and rhotics DA URUDA a

da dab6

NE (b) TE

de3 te

PI (b) b

/de/ (?)

NE (c) NI (a) EN LI

ne ne2, ni en le, li

TI

di3, ti TU DU (a) DU8 c

TA d AD NA

ta ad na

AN (b) LA LA2 AL DU (c) e RA f

an IN in la NI (b) li2 la2(/laH/?) NE (d) li9 al ra2 = /řa/ DU (d) e re6 = /ře/ ra (ru?) URU re2, ri2

du2 du, tu3, (tum2) du8

NU

nu

LU LUL UL

lu lu5(luH?) ul

RU ru UR ur UR3 g ur3

Notes: a URUDA = dab6 renders in Lagaš the combination of two verbal morphemes in the verbal prefijix *da+b. b PI is used for the combination of verbal morphemes *ene+da “with them”. It is difffijicult to reconstruct the pronunciation of the sign in this function. In Old-Babylonian Sumerian the combination is written en.de322 and in Ur III Akkadian the reading te11, ṭi6 is attested (Hilgert 2002, p. 234+22; 375). I therefore tentatively reconstruct the reading as /de/. c DU8 in šu.du8.ur3 (Ukg 4 V 6 ) is phonographical for TUG2.MAḪ = šutur a garment. DU8 and UR3 do not occur very frequently in phonographical spellings, and here they may indicate vowel length or an o-colouring influenced by the following /r/. d TA = ta is used exclusively as the ablative postposition. e DU: ra2 = /ř+a/ and re6 = /ř+e/ are both written with DU.

22  ba.en/e.de3.kur2 “. . were removed from them” Curse of Akkad 99 and he2.en.de3.ĝa2. ĝa2.ĝa2 Šulgi Hymn E 62 ”.. one must place everything beside them.”.

198

theo j.h. krispijn

f RA: The verbal form gur.ra.da.am6 (Ean 1 XVII 10 etc.) “(If) he is someone, who will turn away from . . .” might be a rendering of gur+*e+d+am the imperfective participle and the copula. g UR3: in DUNur3 = /dur/ “stallion” and šu.du8.ur3 (Ukg 4 V 6) possibly for /ūr/ or /or/. Endings:

C+a (1)

C+a (2)

C+e (1)

C+e (2)

DA / -D.A h NA DU = /řa/

DA /-D.∅ NA / -N.∅ DU = /řa/

NE (= de3) NI / -N.E DU = /ře/

NE

RA/ -R.A -R.∅ i

RA/ -R.A*-R.∅

IRI = re2

C+u (1)

DU = /ře/ / -Ř.E

C+u (2)

-Ř.∅

h -D.A: pad.a “weep (with joy)!” Ean 1 XI 18 is an imperative with the verbal morpheme *a, where the prefijix i/a is placed after the verb. This form is one of the indications that this prefijix is in fact /Hi/ or /Ha/.23 i -R.A: i3.ir.a Fö No 109 I 1 etc.

Palatals and alveolars ZA

za

ŠA DU (b)

ša ša4

SA DI e AŠ2 h

sa sa2 aš2

AB2.ŠA3.GI a

ŠE3 (a) ŠE ŠE3 (b) d

zex

še še eš2

ZI

zi/

ZU

zu/

ŠI

ši

KA b KAxUD c ŠU

zu2 zu9 šu

SI SUM f

si si3

SU SU3 g

su su3

Notes: a AB.ŠA3.GI = zex in ha.ni.gaz.zex, an extraordinary variant of the ha.ni.gaz.e (imperfective base) in Ent 28 VI 29 // 29 VI 40. b KA: = zu2 in zu2.lum “date”. c KAxUD: = zu9 in zu9.bar “bronze”. d ŠE3 = eš2 is used as pronominal ending for 3 Plur. “they”. e DI: = sa2 in ma.sa2 a basket DP 51 I 1 etc. g SU3 is used in the alternative spelling nu.da.su3.su3.da.ka (Ent 28 III 27) instead of nu.da.su.su.da.ka. h AŠ2: in Ti.ra.aš2, the name of a temple. f SUM: = si3 in ni.si3.ga (Ean 1 r. I 4) = niĝ2.sig7.ga “the yellow thing”.

23  Verbal forms from later texts corrobating the reconstruction /Hi/ are: izi i3.im.ta.la2 Gudea St F III 2; šu.ni i.im.[ Enmerkar and the Lord of Aratta 170; ugu.na i.im.šu2 Enmerkar and the Lord of Aratta 578; ĝišma2 i.im.da.gid2.da Curse of Akkade A 45.

writing semitic with cuneiform script

199

Endings:

C+a (1)

C+a (2)

C+e (1)

-Š.A / Š.ŠA4 i

-Š.A / Š.∅ -Š.SA -Š.SA2

ŠE3 / -Š.E j

-Z.ZA

C+e (2)

C+u (1)

C+u (2)

-Z.E / -Z.ZEx k

i -Š.ŠA4: DU = ša4: tuš.ša4 DP 418 VIII 7. j -ŠE3: ĝiš.tešx.še3 la2 Ent 28 III 10 // ĝiš.tešx.e la2 Ean 1 IX 1 ” to clash”. (tešx = UR.UR). k -Z.ZEx: see AB.ŠA3.GI above.

Velars, uvulars, gutturals, and pharyngals GA

ga

KA (b)

ka

GI a

ge, gi

KI

ki

GU

gu = /ku/

GIL GIN2 b

gi16 gi11

GI4 a

gi4

KA (a)

gu3

KID*

ke4

GU2 KU

gu2 ku

EZENx BAD c

ubx

AG ĜA2 (a)

ĝa2

ĜA2 (b)

ĝe26

MU (b)

ĝu10

ḪA (a)

ḫa

ḪI*

ḫe, ḫi

ḪA (b) d

ḫa = /ḫu/

GAN

ḫe2

ḪU*

ḫu

a b c d

GI4 changes with GI: si.ge4.de3/da DP 199 I 3 otherwise si.ge.da. GIN2 is attested in the name dGIŠ.NE.GIN2.MES = dBil3.gi11.meš the divine Gilgameš.24 EZENxBAD: = ubx Lagaš-dialect for ug5. ḪA: = /ḫu/ in ĝišḪA.lu.up2 = ḫuluppum a type of oak (Akkadian) DP 75 r. II 5 etc. ḪA.mu- is later written ḫu.mu-: ḫa.mu.AK.e Ukg 15 I 3 (Lit); ḫa.mu.tumu3 Ukg 6 IV 3’ etc. ḪU occurs in ḫuKU.BU = /ḫurum/ DP 573 VIII 3.

24

 See George 2003: 75.

200

theo j.h. krispijn

Endings:

C+a (1)

C+a (2)

C+e (1)

C+e (2)

C+u (1)

C+u (2)

-G.GA

-G.GA / -G.∅

-G.GE

-G.GE / -G.∅

-G.GU e

-G.∅ e

-Ĝ.ĜA2

-Ĝ.ĜA2 / -Ĝ.∅

-Ĝ.ĜE26

-Ĝ.ĜE26 / -Ĝ.∅ e

-H.HA / -H.A f

-H.∅

-H.HE

-H.E / -H.∅ e

e For -G/-Ĝ/-H.∅ cf. the imperfective (marû) forms: gaba mu.ru.da. Ean 5 III 3–5 (= /ruguda/ < ri/ru); zaĝ3.de3 Nik 91 I 1, zaĝ3.zaĝ3.dam TSA 28 II 4 (= /zaĝade/), lah5. da.ke4 DP 135 XV 2 (= /laḫadake/). f -H.A: niĝ2.zu2(ḫ).a Ukg XII 17. zu2 “to steal” is later written zu2.uḫ.

Vowels A

a

E

e

NI

i3

U2 a U3

u2, wax, wux u3

a U2: = wax in dba.wax = wux in ḫa.wux.da VAT 4856 II 7 later ḫa.bu3.da “hoe”. Endings:

V+a (1)

V+a (2)

V+e (1)

V+e (2)

V+u (1)

-A.A

-A.∅

-A.∅ b

–

–

-E.A -I.A e -U.A -Y.∅

-E.∅

-A.E a / -A.∅ c -E.∅ d

-E.∅

–

–

-U.∅ –

-U.∅ f -Y.E h

-U.E g –

-U.∅* –

-U.∅ –

a b c d e f g h

V+u (2)

a.e Fö 100 III 2. a.e < /a.eš2/ (< a+še3) Ent 28 II 35; i3.ra2.ra2.e Fö 109 I 3; -ba.e Fö 173 I 4. ba.(+a < e).me Fö 173 X 2; ra2.(+a < e).de3 Fö 131 IV 3. ugula.(e) Fö 16 III 3. dNanše.(e) Ent 26 12; 35 III 7; en- dNanše.(e)-mu.tu (PN) Fö 132 IV 1; lugal-muš3.me.(e)ki-aĝ2 (PN) DP 220 III 5. No combinations with -I are attested. lu2.(+u < e) Ukg 6 II 15’; Ean 1 IX 3; etc.; en-abzu.(u < e)-si (PN) TSA III 2. -gu7.(+u < e) Ean XVI 24; etc.; -gu7.(+u < e).ne RTC 58 II 1; DP 222 II 8. -gu7.e En I 29VIII 6; Ean 1 VI 15; DP 149 I 7. E2 = /hay/ e2.e-gu2.dam Fö 63 III 4; e2.e-a.na-AK (PN) Fö 63 III 4; nin-e2.ni.še3 (PN) DP 75 VII 3.

writing semitic with cuneiform script

201

Phonograms attested as verbal logograms or morphemes causing e- or i-colouring Verbs -> e Verbs -> e/i Verbs -> i Prefijixes -> e Prefijixes + e Prefijixes + e/i Prefijixes + i

BA, DAB6, DI = sa2, ĜA2, ḪA.(LA), LA2, RA, RU AK, ĜA2, GI4, ḪA.(LAM) BI = be2, DU, DU8, E, LU5, SI, SU, SU3, TI, TU, UD.DU = e3, UŠ, ZI DA, DAB6, MA, NA, NE, PI = /de/ (?). BI=be2, E, ME, ŠE3 MI, NI = ni/ne2 NE = bi2 a, NI = i3, ŠI, TI

Notes: a It is remarkable that the verbal base *e “to speak” (imperfective base) causes i-colouring. The only reason for this would be if the verbal base were pronounced as /ye/. If that were so, then the pronominal ending after the imperfective base will also be reconstructed as /ye/. It is perfectly plausible to take this ending as the (auxiliary) verb e + pronominal ending.

The Syllabary of Ebla Akkadian25 Ebla Akkadian (EbAkk) The royal archives of the city-state of Ebla were fijirst excavated by Paolo Matthiae and his Italian team in 1975 at modern Tell Mardikh in the northwest of Syria. After some early rather speculative publications it became clear that Ebla was the capital of a well organized local state in the 25th century BC, with defijined borders but wider contacts. A local form of Akkadian was used there for administrative, lexical and literary texts. Some of the lexical texts are bilingual. The monolingual Sumerian lexical tradition fijits closely to the Southern Mesopotamian tradition, the oldest traces of which go back as far as the Uruk IV period (3200 BC). The bilingual lexical texts, Sumerian and EbAkk, exhibit a local development from the monolingual lexical lists. These bilingual texts, the largest of which is the ‘Vocabulary of Ebla’ (VE), were primary sources for reconstructing EbAkk. Orthographic conventions The syllabary of EbAkk largely resembles that of SargAkk, being a mixed system of phonograms and logograms. As in SargAkk phonetically writings usually render all the phonemes of a word, but do not denote special features such as consonant doubling and vowel length. 25

 See Rubio 2006.

202

theo j.h. krispijn

EbAkk difffers from SargAkk in writing closed syllables. A C1VC2 syllable may be written as: – C1V1 + V1C2: tu3-uš-ta2-ki2-lum = tuštakkilum “to allow to eat constantly”. – C1V1C2: kal2-la-tum = kallatum “bride, daughter in law” VE 322. – C1V1 + C2V1: The second vowel repeats the vowel of the preceding syllable and is technically redundant: i3-ri2-sa-tum = iršatum “wish” VE 234. – C1V1 + C2V2: The second vowel repeats the vowel of the following syllable and is technically redundant: a-ka3-lu bu3-ṭa-ma-tim = akalu buṭmātim “pistachio-nut bread” VE 32. – C1V: sa-ma = šalmā “be healthy” ( greeting) VE 827. – with a CV1–C+I-sign: ur-ba-tum/u3-ri2-ba-tum/u9-ri2-ba-tum = ḥurbatum “sword” VE 548. Long vowels are written in a few cases with scriptio plena, when E2, I and PI indicate respectively /ā/, /ī/ and /ū/: – E2 = /ā/: a-ba-e2-su-ma-a = abāšumaj (?) “two fathers of both of them ” ARET 11 nr 1 VII 8. – I = /ī/: an.ru12.ru12 = du-uš-da-i-i-lum VE 801 // u3.ru12 = tiš2-da-i-lum VE 1133 = tuštayīlum (?) < tuštaḥḥilum < *tuštanḥilum “to let someone possess permanently”. – PI = /ū/: saĝ.uru5 = a-ša-šu-wu-tum / a-ša-šu-tum = *aṯa(m)ṯūtum “hurricane” VE 242; ḫi.mu.DU = su-lu-wu-um = šūruwum or šūrûm “to lead, to bring” VE 1340; nu.uš = a-wu-um, lu-wu-um = yau(m) < lau(m), lû(m) “surely” VE 1439’; u3-ḫu-wu-tum =*uḫūtum “brotherhood” St.Ebl. 4 174. The phonograms: consonants The orthography of the offfijicial Akkadian of Ebla includes dialectal features which penetrated offfijicial EbAkk spellings: – /l/ was pronounced as /y/ in intervocalic position, so spellings of /l/ with the L-series can be interpreted as offfijicial orthography and those without L as reflecting the pronunciation of the EbAkk dialect.26 This explains why /l/ is often not written.

26  A special case is A-LI-(A) possibly for /ayyu/: al.uš2 = a-li-a-mu-du ayyu(m) mūtu “everyone who is dead” VE 985; ba4.nu11 = ga-nu a-li-ga-na-a qanu(m) ayyu(m) qanay “everyone who is surely jealous” VE 1052.

writing semitic with cuneiform script

203

– The correct (Mesopotamian?) pronunciation of the L-series must have sounded similar to that of the R-series, which explains why the L-series is also used for /r/. – The combination /b+u/ was apparently pronounced as /wu/ or /’u/ in intervocalic position and was written with NI. – /ḏ/ had not coalesced with /z/ in Ebla but it had in SargAkk. The phonemes /ṯ/ and /ḏ/ are both written with the Š-series. – EbAkk distinguished velars and uvulars diffferently from SargAkk. The phonogram for /h/ and /ḥ/ contrasted with the phonogram for /c/ and /’/: E2 = /ha/, /ḥa/ :: A = /’a/, /ca/; U9 = /hu/, /ḥu/ :: U3 = /’u/, /cu/. Processes of convergence diffferent from those at Ebla occurred in Mesopotamian Akkadian after the Sargonic period: the convergence of /’/ with /h/ had no influence on the vowel /a/, but /c/ converging with /ḥ/ produced a > e. – Some phonograms, such as UD, EN, IR and representing /h/+V or /ḥ/+V may also represent /y/+V. The phonograms: vowels The vowels of EbAkk also show features diffferent from SargAkk: – Apparently EbAkk made no distinction in writing between phonograms with /e/ or /i/, for E seems to render /’i/ (see below, table on labials). – The diphthongs /ay/ and /aw/ are always contracted in SargAkk into /e/ and presumably /o/ (written with Cu or uC signs),27 respectively, but not always in EbAkk (see below, table on semi-vowels). – It is possible that /‘a/ > /i/ in some cases (cf. note to IB2). – A distinction is made between short ‘closed’ and long ‘open’ vowels. i.e., whatever consonant follows, the vowel indicated by some phonograms is always short: DA5 = /daH/, GA2 = ba4 = /baH/, MA2 = /maH/.

27

 See the section on Sargonic Akkadian vowels.

204

theo j.h. krispijn

Tables of the EbAkk phonograms Labials C+A/ A+C

phonogr.

C+I/ I+C

phonogr.

C+U/ U+C

phonogr.

BA GA2 (BAD a PEŠ2 AB AB2 IB2 b MA, MA2 AM AN

ba, pa ba4, /baH/ /baH/) /ba/ ab, ap /Hab/ /‘ab/ ma /maH/ am am6

NE PI

bi2, pi2 bi, pi

IB IB2

ib, ip ib2

BU NI BU3 KA UB

bu, pu, /bu/, /wu/ bu3, pu3 /bu/ ub, up

MI ME IM

mi mi3 im

MU

mu

UM UD c

um /Hum/, /yum/

Notes: a BAD is used only in compounds with ba’lum, and can therefore be considered as logographic writing, as it is in SargAkk. b IB2 = /‘ab/ in IB2-dum = /‘abdum/ “servant, slave” VE 253; IB2-IB2-ba-tum = /‘ap‘appatum/ “eyelid” VE 1204 (/‘a/ > /i/ ?). c UD = /Hum/ or /yum/.

Dentals, interdentals, laterals and rhotics DA URUDA

da, ta2, ṭa da5, /daH/

AD NA

ad, at na

AN

an

ŠA a

/ṯa/, /ḏa/

AŠ

/aṯ/, /aḏ/

LA c

la, /ra/

TI NE

di3, ti, ṭi3 ti9

TE DI

ti4 di, ṭi

NI NE IN EN ŠI ŠE3 IŠ11

ni ni5 in /yin/, /Hin/ /ṯi/, /ḏi/ /ṯi//, /ḏi/ /iṯ/, /yiṯ/ /iḏ/ iš12 = /iṯ/, /yiṯ/ li, /ri/ li2, /ri/ li9

LAM LI NI b NE

DU TU

du, tu2, ṭu3 du2

UD NU

ud, ut nu

UN ŠU ZU UŠ

/ṯu/, /ḏu/ /ḏu/* /uṯ/, /uḏ/

LU LU2 LUL

lu, /ru/ lu2 lu5

writing semitic with cuneiform script

205

Table (cont.) AL6

al6, /ar/

RA

ra

AR

ar

IL IL2 URU

il il2, /yil/, /yir/ ri2

IR IR3

ir, /yir/, /Hir/ ir3, /yir/, /yer/

UL

ul, /ur/

RU EN UR

ru ru12 ur

Notes: a ŠA alternates with ZA in consonant clusters: gu2-da-ša/za-tum = qudaštum (?) VE 159, ma-za/ša-lu-(um) VE 615 en ŠA/SA: mu-sa-da-tum/en-ni-ša-du VE 615 et passim in divine names. b NI = li2 (against Rubio 2006) is attested in li2-gi-da-tum = liqtātum VE 525 // li-gi-tum = liqtum VE 282b. c L for R: ra-a-mu-um / la-a-mu-um = ra’āmum “to love, to reconcile” VE 237; 1208.

Palatals and alveolars ZA

za, sa3, ṣa

AZ SA DI

az, as, aṣ sa = /ša/ sa2 = /ša/

AŠ2

as2 = /aš/

ZI ZE2 GIŠ SI SE11 ŠE3 IŠ LAM

zi, si2, ṣi2 ṣi is (?) si = /ši/ si11 = /ši/ /ši/ is2 = /iš/, /yiš/ iš12

ZU KA

zu, /ḏu/ su2, ṣu2 zu2, /ṣu/

SU SU3*

su = /šu/ su3= šu

UŠ

us2 = /uš/

Notes: a For ZU = /ḏu/ cf. a-ha-zu-um VE 215 // a-ha-šum2 VE 1284. = /ahāḏum/. b SU3 = /šu/ only as pronominal sufffijix.

Velars, uvulars, gutturals, and pharyngals GA

ga, ka, qa2

AG ḪA

ag, ak, aq ḫa

UḪ Aa

aḫ /’a/, /ca/, /ha/

NI E2 c

’a5, /ca/ /ḥa/, /ha/*

GI KI GI4 IG ḪI GAN UḪ NI Ec

gi, ki2, qi3 ki ki ig, ik, iq ḫi ḫi2 iḫ i3 = /’i/, /ci/, /’i/ ?

GU GU2 KU UG ḪU

gu, ku8, qu3 gu2, /ku/, qu6 ku uq ḫu

UḪ U3 b

uḫ u3 = /’u/, /cu/

U2 U9 e

/’u/, /yu/ /hu/, /ḥu/

206

theo j.h. krispijn

Notes: a A = /’a/, /ca/: a-na-a = canay “of two eyes” VE 712; 715, 723; 728. b E2 = /ha/, /ḥa/: e2-a-gu2-um / a-la-gum2 = halākum “to go” VE 984; e2-ba-lu = ḥablum “rope” VE 1174. c E is seldom used, and then only in personal names: il2-a/e-i-sar = yil’a/i-jišar “Yišar prevailed” TM 76.G.32. d U3 = /’u/, /cu/: a-a-u3-mi/i3-a-u3-um = yā’um “mine” VE 616; la-u3-um = rācum “friend” VE 1061. e U9 = /hu/, /ḥu/: u9-ru12-um = hūrum “swamp” VE 438; qa2-ma-u9-um = qamāḥum “to grind” VE 169.

Semi-vowels Aa A-A NI CA-A c PI A-PI d

/ya/, /ay/ /ya/ /ya/, /ay/ /Cay/ wa /aw/

I

/yi/

U9 b A-LI

/yu/ /ayyu/*

PI

wi

PI

wu

U3 NI e

/wu/ /wu/

Notes: a A = /ya/ sometimes for /la/: di = a5-a-ga-du-um = /ayaktum/ < alaktum. Cf. also la-i-mu / a-i-mu = laḥ(i)mum “packed together” VE 63; na-pa2-u9 = napāyum < napālum VE 325. b U9 = /yu/: e2-u9 = ḥayyum “the god Ea” VE 803. c CA-A = /Cay/: na-si-i3 a-na-a = naši’ canay “lifting of the two eyes” VE 723. d /aw/ is sometimes not contracted: a-wu-um, lu-wu-um = /lau(m)/, /lû(m)/ “surely” VE 1439’ e NI = /wu/ for /bu/: ki2-ṣi2-ri2 bu3-dim / NI-dim = /wudim/ = kiṣir būdim “shoulder joint” VE 1210.

Sumerian in Ebla (EbSum) The early Sumerian lexical tradition, which formed the basis for the education of scribes in Mesopotamia and also in Syria, is fijirst seen in the Uruk III Period, ± 3000 BC. That is why most lexical texts found in Ebla belong to the same series of texts found in sites such as Fara and Abu Ṣālābīkh in Mesopotamia. But the scribes of Ebla used phonographical versions of these lexical series in their schools next to the orthographically written series. The syllabary of the phonographical versions resembles the one of EbAkk. Apparently they have developed these lists to learn how to pronounce Sumerian words.28 28  First publication Pettinato 1981. The following texts are identifijied: MEE 3 No 61 (phonographic) = translation of Wordlist B ‘Stone’ on pp. 143–155; No 62 = Animal List A

writing semitic with cuneiform script

207

Young scribes were instructed in Akkadian, the offfijicial language of Ebla. The ‘Vocabulary of Ebla’ (VE) is the most important representative of a bilingual series, which begins to crystallize the process of this Akkadian instruction. The other bilingual lexical lists do not belong to one series. Rather they are ‘Estratti di Vocabolari’ (EV), excerpts related to VE and lists composed for a more limited practical purpose, and they sometimes contain phonographical glosses to indicate the pronunciation of Sumerian words.29 Early in the curriculum of Ebla a young scribe would learn to read a special syllabary of Sumerian signs together with the names of those signs in Akkadian, i.e., how to pronounce the sign with an ending -um. This selection of Sumerian logograms acted as a bridge to more complicated lists, such as the list of professions ‘Early Dynastic Lú A’. The Sumerian literary tradition (see Edzard 1984), as known from Fara and Abu Ṣālābīkh, is poorly represented in Ebla.30 We have two myths about the sun god Šamaš and the goddess of scribal art Nisaba respectively which are written in an early form of Akkadian. The fijirst text is reconstructed from broken tablets found at Ebla and Abu Ṣālābīkh.31 By contrast we have many Sumerian incantations from Ebla in ‘orthographic’ and phonographic versions. These literary texts are not included in this survey. The repertoire of signs in the early curriculum, i.e., the Ebla Syllabary (Tables A), is more limited than that of works taught in the later curriculum, such as the thematic lists and the excerpts (Tables B), as becomes clear from the tables below (see tables below). Orthographic conventions The phonographic rendering of Sumerian words in the lexical lists generally follows the orthography of EbAkk (see orthographic conventions discussed earlier).

(domestic animals) on p. 47–56; No 63 = Wordlist D on pp. 165–175. No 64 = Fish list on pp. 91–104; No 68 // 70 // 71 List of parts of the body (partly phonographic) // IAS No 523 from Abu Ṣālābīkh (Postgate, N. Iraq 40 (1978), 101–117). 29  First publication Pettinato 1982. The following EV texts contain glosses: c (excerpt from i), h, i, q. The practical lists refered to are: List of Names (ARET 5 No 24, 25, 26) and List of Numbers (MEE 3 No 54). 30  Two versions of one mythological text about the god Amaušumgalanna ARET 5 No 20–21 with a parallel in Abu Ṣālābīkh (IAS No 278). 31  ARET 5, No 6 // IAS No 326, 342 and ARET 5 No 7.

208

theo j.h. krispijn

Closed syllables follow EbAkk norms with: – C1V1 + V1C2: ŠITA: ti-iš-da-num2 = tištan-um “prayer priest” Ebla Syll. 1. – C1V1C2: ĜIŠGAL: ĝiš-gal-lum = ĝišgal-um “station” Ebla Syll. 15. – C1V1-C2V1: IDIGNA: (’)i3-ti-gi-ra-um = (’)idigra-um (for idigla-um) “River Tigris” Ebla Syll. 34. – C1V1-C2V2: ADKIN = (’)a-ti-gi-num2 = (’)adkin-um “piece of meat” Ebla Syll. 13 – C1V = C1VC2: ab2.kun.babbar = ab-kum-bu3-ru12-um = abkuĝbubr-um “cow with a white tail” (babbar > bubr-um) Animals A 18. – CV1-C+I-signs: LAHAR: na-hi-ru12-um = naḫr-um (for laḫr-um) “ewe” Ebla Syll. 64.32 Phonographic Sumerian lists also use scriptio plena: – /ā/ with E2: TAKA4.ALAN = la-e2-num2 = (a)lān-um “image, statue” Ebla Syll. 52. – /ī/ with I: KID = gi-i-dum = kīt-um “reed mat” Ebla Syll. 56. – /ū/ with PI = wu: DU5 = du-wu-tum = dūd-um “hill, tell” Ebla Syll. 22; with u3: KAxKID.DU3 = su-mu-du-u3 = /šumudû/ h IV 5–6. The phonograms: consonants EbSum does not use signs such as PA, TA and KA, so that voiced :: voiceless oppositions, such as b :: p, d :: t and g :: k are obscured.33 Notes on individual consonant phonemes: – /ĝ/ is mostly rendered fijinally by GA, GI or GU , initially by ME, MU, NU, and medially by NA.

32  Since there was a shift l > y in Ebla-Akkadian, phonograms with n were used to render the Sumerian l. Another example is nu-gal = lugal “king”, passim in ARET 5 No 20; 21. 33  Cf. Old Babylonian series ProtoEa:

Sign

Ebla Syllabary

Line

Proto Ea

Line

PIRIĜ PEŠ2 LAḪTAN KISAL

ba-ri2-gum2 bi2-sum la-ḫa-da- gi-za-lum

87 86 132 9

pi-ri-ig pe-eš la-aḫ-ta-an ki-sa-al

572 583 821 220

writing semitic with cuneiform script

209

– n in Ebla = l in Proto-Ea: KALAM = gi-ne-mu-um = kinim-um Ebla Syll. 45. Initial /n/ is often deleted, as it is later: NINDA = en-da-um = indaum Ebla Syll. 80. – n in Ebla = m in Proto-Ea: URUDA = ne-ba-la-ga-um = nebalaga-um // min-bu-lu-ug = BULUG/BULUG Proto-Ea 804. – /r/ is written with l: SIG7: u3-ga-lum = ugar-um “meadow” Ebla Syll. 24. – /h/ is attested in UR3: u9-ru12-um = hur-um Ebla Syll. 57; GUR8: maru9-um = marh-um “storm” Ebla Syll. 57; MUNSUB-šessig = su-du-u9-um = sut(u)h-um “?” Ebla Syll. 60; UMUN2: nu-u9-mu-um = nuh(u)m-um “mold” Ebla Syll. 12. – EbSum diffferentiates the sibilants: s :: š. 1. s sometimes renders later š in: ŠITA3: mi3-si-(za-gum2) = ĝiš(s(i)takkum) Ebla Syll. 6 // ĝe6-eš Proto-Ea 641.34 2. š sometimes renders z in Proto-Ea: SUHUR: šu-hu-ru12-um = šuḫ (u)r-um Ebla Syll. 31. // zu-ḫu-ur Proto Ea 809. The phonograms: vowels Proto-Ea diffferentiates between words with e and words with i, but the Ebla syllabary does not: PEŠ2 = bi2-šum2 = piš-um “mouse” Ebla Syll 86 // pe-eš = PEŠ2 = /peš/ Proto-Ea 583 :: EZEN35 = i3-zi-num2 = isin-um Ebla Syll. 29. // EZEN = i-zi-in Proto-Ea 759. Sumerian words in Ebla sometimes have a diffferent vowel pattern from later spellings, such as Proto-Ea. These examples from the Ebla Syllabary give an overall picture: – Ebla Sum. /a-a/ = Proto-Ea /i-a/: e.g., SILA4: sa-la-um = šala-um Ebla Syll. 9 // si-la = SILA4 “lamb” Proto-Ea 84. – Ebla Sum. /a-a/ = Proto-Ea /u-u/: URUDA = ne-ba-la-ga-um = nebalagaum // min-bu-lu-ug = BULUG/BULUG “double pin” Proto-Ea 804. – Ebla Sum. /a-i/ = Proto-Ea /i-i/: e.g., PIRIĜ: ba-ri2-gum2 = par(i)ĝ-um Ebla Syll. 9 // p[i-ri-i]g = PIRIĜ “lion” Proto-Ea 572. It is not excluded that Ebla a-i reflects Sumerian e-e.

34

 z renders /š+t/: in: ŠITA3: mi3-si-za-gum2 = ĝišš(i)tak-um) Ebla Syll. 6 // ĝeš-(šita-ak) Proto-Ea 641. 35  The name of this sign EZEN is based on the pronunciation glosses of this sign in syllabaries from the fijirst millennium BC (Sa 383).

210

theo j.h. krispijn

– Ebla Sum. /u-u/: ANŠU = u3-šu-wu-um = uššuw-um or uššûm “donkey” Ebla Syllabarium 85 // ANŠU = an-šu Proto-Ea 566. – Ebla Sum /i/ = Proto-Ea /u/: e.g., BUR2: bi2-ri2-LUM (= /yum/) = biriy-um Ebla Syll. 37. // bu-ur = BUR2 “to release” Proto-Ea 800. – Ebla Sum /i-a/ = Proto-Ea /i-i/: e.g., ŠILIG: si-la-ga-um = šil(a)ga-um Ebla Syll. 14 // si-li-ig = ŠILIG “mighty” Proto-Ea 548a. – Ebla Sum /i-i/ = Proto-Ea /a-a/: e.g., KALAM: gi-ne-mu-um = kinim-um Ebla Syll. 45. // ka-lam = KALAM “nation” Proto-Ea 596. – Ebla Sum /u/ = Proto-Ea /a/: e.g., LAGAR = nu-gu2-lum = nug(u)r-um Ebla Syll. 38. // na/la-gal = LAGAR a priest Proto-Ea 165. – Ebla Sum /u/ = Proto-Ea /e/: e.g., EREŠ: u3-ru12-sum = ur(u)š-um Ebla Syll. 30 // e-re-eš = NIN “queen” Proto-Ea 420. – Ebla Sum /u-a/ = Proto-Ea /i-a/: e.g., WlD 50: LAK350.sila3.aš.gal = u3-šu-la-ga-la Word List D 50. In some of these sign names the second vowel should possibly be omitted, e.g., si-la-ga-um = šilga-um instead of šilaga-um. The diphtongs /aw/ and /ay/ are attested in EbSum, but are not known from Sumerian in Mesopotamia. In EbAkk diphtongs were not always contracted and so constitute an alternative pronunciation also in Sumerian. As in the Sumerian of Mesopotamia initial vowels can be deleted: TAKA4.ALAN = la-e2-num2 = (a)lānum “image, statue” Ebla Syll. 52; AMAR = ma-e2-ru12-um = (a)mārum “calf ” Ebla Syll. 74. Tables of the EbSum phonograms The tables below contain CV, VC, C+UM phonograms. The C+UM signs are added because they render the fijinal consonant followed by the Akkadian nominative ending -um in the syllabary. The fijirst table (A) contains phonograms occurring only in the Ebla Syllabary. The second table (B) contains phonograms known from other sources such as: – The phonographical version of the Animal List A (MEE 3, No 62 pp. 251– 252 // orthographical version pp. 47–56).

writing semitic with cuneiform script

211

– The phonographical version of the Word List D (MEE 3, No 63 pp. 252– 254 // orthographical version pp. 165–175). – The phonographical version of the Fish List A entries from 33–67 (MEE 3: No 64 p. 254// orthographical version pp. 105–120) (See also Archi 1992: 7). – The phonographical glosses in List of Bodyparts (MEE 3: No 68). – The phonographical glosses to the Ebla word lists labelled by Pettinato as ‘Estratti di Vocabolari’ (EV), that do not belong to the big lexical series VE: “Abstracts of Vocabularies” text c (MEE 4: 93–95); text f (MEE 4: 97); text h (MEE 4: 99–100); text i (MEE 4: 100–103); text q (MEE 4: 105–106). Labials A C+A / A+C

C+I / I+C

C+U / U+C

BA MA

NE = bi2 ME = mi3 IM

BU3 = bu3, bum MU UM

NE = bi2

BU3 UB

ME = mi3 MI IM

MU

B BA AB AB2 MA, MA2 AM

UM

Dentals, interdentals, laterals and rhotics A DA

TI NE = ti9

NA

NE = ni5

ŠA

EN = in4 ŠE3 = ši4

LA (A = ya < la) RA

LI NI=li2 URU = ri2

DU TU IB2 = tum NU LUM = num2 ŠU SUM = šum2 LU LUM EN = ru12

212

theo j.h. krispijn

B DA DAB6 NA AN ŠA ŠA3

TI NE = de3, ti9 NE = ti9 IN EN = in4 ŠE3 = ši4

LA

IŜ11 LI

AL RA AR

URU = ri2 IR

DU TU IB2 = tum NU UN LUM = num2 ŠU LU LUM UL EN = ru12 UR UR3

Palatals and alveolars A ZA SA

ZI SI

AŠ2

IŠ

ZU SU ŠUM = sum6 UŠ

B ZA a SA

ZI SI

AŠ2

IŠ

ZU SU b ŠUM = sum6 UŠ

Notes: a ZA stood apparently for the cluster /št/ or /tš/ in ŠITA3: mi3-si-za-gum2 = ĝišs(i)tak-um (?) Ebla Syll. 6. b SU sometimes renders /zu/: AZUe-su.HUB2.DU = eAZU-kas-du MEE 4: 97 f 1.

writing semitic with cuneiform script

213

Velars, uvulars, gutturals, and pharyngals A GA

GI b

ḪI

GU2 c LUM = gum2 d GUR GU2 c = /ĝu/ LUM d = /ĝum/ MU f = ĝu10 (initial) NU g = /ĝu/ (initial) ḪU

GAL GA a = /ĝa/

GI b = /ĝi/

NI = i3

U3

ME e = /ĝi/ (initial) ḪA ḪAR A E2 = /ha/ NI = /a/

U9 = /hu/

Notes: g = ĝ: a SAĜA: um-bi2-sa-ga-im = umbisaĝa+um (?) Ebla Syll. 16; 63. NAĜA: na-ga-um = /naĝa+um/ Ebla Syll. 63. b NIMGIR: ne-gi-ru12-um = niĝir+um Ebla Syll.; NESAĜ: li-sa-gum2 = nesaĝ+um Ebla Syll. 10. c KUN: gu2-gu2-um = kuĝ+um Ebla Syll. 59; ŜEĜ9: še3-gu2-um = šeĝ+um Ebla Syll. 95. d PIRIĜ: ba-ri2-gum2 = par(i)ĝ+um Ebla Syll. 87. m/n = ĝ: e ŠITA3: mi3-si-za-gum2 = ĝišš(i)tak+um (?) Ebla Syll. 6; f ĜISAL = mu-š[a-lum] = ĝušal+um (?) Ebla Syll. 125’’; g ĜURUŠ = nu-ri2-šum2 = ĝur(i)š+um Ebla Syll. 46.

B GA

ĜA2 ḪA A E2 a = /ha/ NI = /a/

GI KI

GU2 GU LUM = gum2 KUM NE = kum2

ḪI NI = i3 E=eb

ḪU U3* U9 = /hu/ U9

Notes: a E2 = /ha/ = /ā/ šu.ta šu.de6 = ga--ma ga-ti-iš (qatumma qatiš) = šu-da-e2 šu-ti-a = šutā šudea = “recieved from the hands” MEE 4: 106 q 35; TAKA4.ALAN = la-e2-num2 = lānum “image, statue” Ebla Syll. 52; AMAR = ma-e2-ru12-um = (a)mārum “calf ” Ebla Syll. 74; DUL3 = lu-ma-e2-šum = lumāš-um Ebla Syll. 49. b E = e (?) in azue-su.HUB2.DU = eazu.kas.du “?” MEE 4: 97 f 1.

214

theo j.h. krispijn

Semi-vowels A A a = /ya/, /ay/

I b = /yi/

U9 c = /yu/ LUM e = /yum/ (?) PI f = wu

B A = /ay/ PI = wa A+PI g = aw

I PI = wu

Notes: a A = /ay/ (?) in GIŠGALxAŠ = sa-da-a = /šaday/ = da-ri2-du-um “?” MEE 4: 97 f 15 (EV). b I = /yi/, plene /ī/: KID = gi-i-dum = kītum “reed mat” Ebla Syll. 56; TIR = ti-i-ru12-um = tīrum “forest” Ebla Syll. 75. c U9 = /yu/ “5” passim in Wordlist D (MEE 3: 165–175). d LUM in word-fijinal position, indicates the sufffijix um in the syllables lum, num2 or gum2. The reading of fijinal LUM is especially problematic as the last syllable of a sign name which, according to later glosses, has no fijinal stem consonant. In these cases probably LUM = /yum/: BUR2 = bi2-ri2-LUM = biri( y)-um (Proto-Ea 800: bu-ur) Ebla Syll. 37; DARA4 = ti9-ri2-LUM = diri( y)-um Ebla Syll. 58; GEME2 = gi-mi3-LUM = gimi( y)-um Ebla Syll. 84. e PI = /wu/, plene /ū/: DU5 = du-wu-tum = dūdum “hill, tell” Ebla Syll. 22; AD6 = mu-wu(?)tum = mūtum “dead (body)” Ebla Syll. 88. f The ending um after -u is written wu-um: TAKA4 = /šu/ = šu-wu-um = šu(w)-um or šûm “hand” (?) Ebla Syllabarium 51; ANŠU = u3-šu-wu-um = uššu(w)-um or uššûm “donkey” Ebla Syllabarium 85. g aw/u: MI+ŠITA (= utu2) = a-wu-zu = /awzu/, /uzu/ “soup” Word list D 41 // MI+ŠITA.I3 = u3-zi-i3-ti-a uzux.i3.de2.a ” soup with oil (poured in)” Word list D 43.

Synthesis The inhabitants of Southern Mesopotamia in the time of Sargon were to some extent bilingual. The two languages used simultaneously in similar types of documents were Sumerian, the original language of the Southern city-states, and Akkadian, the language of the conquerors, of military personnel and of those who had become aristocrats. The same script was used for both languages but they followed separate scribal traditions. The orthography of Sargonic Akkadian stems from the north-eastern area of South Mesopotamia, but that of Sumerian maintained the offfijicial tradition of southern city states such as Umma, Lagaš, and Ur prior to the language reforms of Narām-Sîn (Krispijn 2005).

writing semitic with cuneiform script

215

There was certainly interaction between the two languages, even though in only relatively few cases is the same scribe found writing texts both in Sumerian and Akkadian. Scribes who had learned to write Sumerian, learned how to spell an increasing amount of Akkadian names in the Sargonic period (Westenholz 1977). Sumerian orthography was subjected to a more complete and systematic way of writing closed syllables after Narām-Sîn’s scribal reforms (cf. Krispijn 2005). But, on the other hand, Sumerian maintained its own spelling conventions and a syllabary diffferent from Akkadian, because Sumerian, apparently, had a distinct phonemic system. There are more and more varied Sumerian documents from the PreSargonic period than from the Sargonic period and constitute a better source for Sumerian orthography than the later documents. Comparing the orthographic norms of PreSargSum with those of SargAkk shows the modifijications to the script introduced by a scribe moving from one language to another. Texts written in EbAkk, slightly earlier than the Sargonic period, are less complex than those from Mesopotamia. At Ebla scribes learned Sumerian and Akkadian side by side from syllabaries and lexical series (including bilingual ones) which had been partly developed locally, and their familiarity with Sumerian literature was more limited. Closed syllables were treated diffferently. In PreSargSum the fijinal consonant was not consistently written. In SargAkk it was, and in Ebla almost always closed syllables were written fully in Akkadian as well as Sumerian. The Syllabary of Ebla, a beginner’s text book, includes fewer phonograms than are found in other lexical lists from Ebla, suggesting some were taught only later in the curriculum. The variations of stops in Sumerian and Akkadian were diffferent. In Sumerian apparently the only distinction was voiceless :: glottalized, whereas in Akkadian there were three distinctions, voiceless :: voiced :: emphatic (i.e., perhaps glottalized). In Akkadian sometimes one phonogram (especially those of the type Ca) was polyvalent for all three distinctions, e.g., DA = /da/, /ta/, /ṭa/. By contrast TA, often found in Sumerian to indicate the postposition ta “from”, is never used in Akkadian and it is unknown how it would have been pronunced there. This applies also to the BV-, ZV-, GV- series. Similarly KA is used only in Sumerian and not in SargAkk or EbAkk. The spelling of endings is special source for the phonology of Sumerian, sometimes resumptive is used, sometimes broken spelling

216

theo j.h. krispijn

The practice of using sometimes a resumptive instead of a broken spelling for Sumerian postpositions and other morphemes shows that the sibilant traditionally transliterated Š was pronounced diffferently in Sumerian and in Akkadian. Sumerian broken spellings show that Š was apparently aspirated or glottalized /čh/, /č’/, whereas in SargAkk it represents /ṯ/. S most probably represented /š/ in both Sumerian and Akkadian. Because Sumerian /ĝ/ and /ř/ were unknown to Akkadian in the phonographical sections of syllabaries and lexical texts from Ebla they were rendered diffferently according to their position in the word. Some other phonemes were also pronounced slightly diffferently, such as /l/, /w/ and /r/. In SargAkk four, or even fijive, vowels are distinguished, a-e-i-(o)-u, but in EbAkk we fijind only three, a-i-u. The phoneme /e/ is originally Sumerian, since the oldest Akkadian did not isolate /e/ as a separate phoneme. The vowel pattern of bisyllabic words seen in the Ebla Syllabary is sometimes very diffferent from reconstructions based on the Old-Babylonian syllabary Proto-Ea. Long vowels were not systematically indicated in PreSargSum, but forms of scriptio plena were developed in SargAkk In Akkadian as well as in Sumerian in the Syllabary and lexical texts of Ebla scriptio plena indicates phonemic vowel length, /u/ is distinguished from /ū/ in SargAkk by using specifijic phonograms. In special cases PreSargSum, SargAkk and EbAkk all distinguish between a short vowel in a closed syllable, where the vowel is followed by a consonant or hiatus, and long vowel in an open syllable, e.g., MA = /mā/ :: MA2 = /maH/, where in Ebla /H/ usually renders one of the laryngeals or the pharyngeals, but in SargAkk H can render Akkadian /y/, /’/, /c/, /h/, or /ḥ /. Later in Southern Mesopotamia, in what became Babylonia, phonograms previously used exclusively for Sumerian in the Sargonic period came to be increasingly used also in Pre-Old-Babylonian (Hilgert 2002) and Old-Babylonian. The phoneme /e/, which had only seldom been distinguished from /i/ in SargAkk, was written more and more systematically in Babylonian of the Post-Sargonic period. Therefore it is reasonable to suppose that the pronunciation of the Babylonian dialect of Akkadian was strongly influenced by the Sumerian substrate.

writing semitic with cuneiform script

217

References36 Archi, A. 1992. Transmission of the Mesopotamian Lexical and Literary Texts. Fronzaroli, P. Quaderni di Semitistica 18: 1–39. Firenze: Dipartimento di Linguistica. Attinger, P. forthcoming. Tableau grammatical du sumérien (problèmes choisis), prepublication on the internet: http://www.arch.unibe.ch/content/ueber_uns/pascal_attinger/ index_ger.html. Bauer J. 1998. Der vorsargonische Abschnitt der mesopotamischen Geschichte. In P. Attinger & M. Wäfler (eds.), Mesopotamien, Späturuk-Zeit und Frühdynastische Zeit, OBO 160/1, pp. 429–585. Freiburg: Universitätsverlag. Borger, R. 2003. Mesopotamisches Zeichenlexikon. Münster: Ugarit-Verlag. Civil, M. 1979. Materials for the Sumerian Lexicon MSL XIV. Roma: Pontifijicium Institutum Biblicum. —— 2007. Early Semitic Loanwords in Sumerian. Roth, M.T. Farber, W. Stolper, M.W. von Bechtolsheim, P. From the Workshop of the Chicago Assyrian Dictionary Volume 2 Studies Presented to Robert D. Biggs. Chicago: Oriental Institute: 11–33. Diakonofff, I.M. 1983. Some Reflections on Numerals in Sumerian. Journal of the American Oriental Society 103: 83–93. Edzard, D.O. 1984. Hymnen, Beschwörungen und Verwantes, Archivi Reali di Ebla Testi (ARET) 5. Rome: Missione archeologica italiana in Siria. Englund, R.K.—Nissen H.J. 1993. Die lexikalische Listen der archaischen Texte aus Uruk, Archaische Texte aus Uruk (ATU) Band 3. Berlin: Gebr. Mann Verlag. Faber, A. 1981. Phonetic Reconstruction. Glossa 15: 233–262. —— 1984. Semitic Sibilants in an Afro-Asiatic Context. Journal of Semitic Studies 29: 189– 224. —— 1985. Akkadian Evidence for Proto-Semitic Afffricates. Journal of Cuneiform Studies 37: 101–107. Gelb, I. 1961. Old Akkadian Writing and Grammar. Chicago: The University of Chicago Press. George, A.R. 2003. The Babylonian Gilgamesh Epic. Introduction, critical edition and cuneiform texts, Volume 1–2. Oxford: Oxford University Press. Gragg, G.B. 1973. Sumerian dimensional infijixes (AOATS 5). Neukirchen-Vluyn: Neukirchener Verlag. Hasselbach, R. 2005. Sargonic Akkadian A Historical and Comparative Study of Syllabic Texts. Wiesbaden: Harrassowitz Verlag. Hilgert, M. 2002. Akkadisch in der Ur III-Zeit (Imgula Band 5). Münster: Rhema. Jagersma, A. 2005. Sound Change in Sumerian: the So-Called /dr/-Phoneme. In J. Black & G. Zólyomi (eds.) Acta Sumerologica ( Japonica) 22 (Special Volume in Honor of Professor Mamuro Yoshikawa), pp. 81–87. Hiroshima. Kloekhorst, A. 2008. Etymological Dictionary of the Hittite Inherited Lexicon. Leiden: E.J. Brill. Kouwenberg, N.J.C. 2003. Evidence for Post-Glottalized Consonants in Assyrian. Journal of Cuneiform Studies 55: 75–86. Krebernik M. 1994. Recension of Green, M.W.—Nissen, H.J. 1987 Zeichenliste der archaischen Keilschriftzeichen aus Uruk (ZATU) Archaische Texte aus Uruk (ATU) Band 2, Berlin: Gebr. Mann Verlag. Orientalistische Literaturzeitung 89: 380–385. —— 1998. Texte aus Fara und Tell Abū Ṣālābīkh. In P. Attinger & M. Wäfler, Mesopotamien, Späturuk-Zeit und Frühdynastische Zeit, OBO 160/1, pp. 235–427. Freiburg: Universitätsverlag.

36  See for abbreviations: See: http://cdli.ucla.edu/wiki/doku.php/abbreviations_for_ assyriology.

218

theo j.h. krispijn

Krispijn Th.J.H. 2005. The change of offfijicial Sumerian in the City-State of Lagash. In J. Black & G. Zólyomi (eds.), Acta Sumerologica ( Japonica) 22 (Special Volume in Honor of Professor Mamuro Yoshikawa), pp. 153–175. Hiroshima. Laroche, E. 1968. Documents en langue hourrite provenant de Ras Shamra. In Ugaritica V, pp. 447–544. Paris Imprimerie nationale-Librairie orientaliste Paul Geuthner. Michalowski, P. 2006. The Lives of the Sumerian Language. In S. Sanders, Margins of Writing, Origins of Cultures, pp. 159–184. Chicago: The Oriental Institute of the University of Chicago. Pettinato, G. 1981. Testi lessicali monolingui della biblioteca L. 2769, (MEE 3). Napoli: Istituto Universitario orientale di Napoli Seminario di studi asiatici. —— 1982. Testi lessicali bilingui della biblioteca L. 2769, (MEE 4). Napoli: Istituto Universitario orientale di Napoli Seminario di studi asiatici. Rubio, G. 2006. Ebla Akkadian, Akkadian, and East Semitic, in Deutscher G.—Kouwenberg N.J.C. (ed.), The Akkadian Language in its Semitic Context Studie in the Akkadian of the Third and Second Millennium BC. Leiden: Nederlands Instituut voor het Nabije Oosten: 110–139. Segert, S. 1988. Die Orthographie der alphabetischen Keilschrifttafeln in akkadischer Sprache aus Ugarit. Studi Epigrafijici ed Linguistici sul Vicino Oriente Antico 5: 189–205. Sollberger, E. 1961. Le syllabaire présargonique de Lagaš. Zeitschrift für Assyriologie 54: 1–50. Sommerfeld W. 2006. Die ältesten semitischen Sprachzeugnisse—eine kritische Bestandaufnahme. In G. Deutscher & N.J.C. Kouwenberg (eds.), The Akkadian Language in its Semitic Context Studie in the Akkadian of the Third and Second Millennium BC, pp. 30–75. Leiden: Nederlands Instituut voor het Nabije Oosten. —— 2010. Prä-Akkadisch. Die Vorläufer der “Sprache von Akkade” in der frühdynastischenZeit. In Kogan L. – Koslova N. – Loesov S. – Tishchenko S. (eds.), Language in the Ancient Near East Proceedings of the 53e Rencontre Assyriologique Internationale Winona Lake: Eisenbrauns. Steiner, R.C. 1982. Afffricated ṣade in the Semitic Languages. New York: The American Academy for Jewish Research. Sürenhagen D. 1999. Untersuchungen zur relativen Chronologie Babyloniens und angrenzender Gebiete von der ausgehenden Ubaidzeit bis zum Beginn der Frühdynastisch-II-Zeit HSAO Band 8. Heidelberg: Heidelberger Orientverlag. Westenholz, A. 1991. The Phoneme /o/ in Akkadian. Zeitschrift für Assyriologie und vorderasiatische Archäologie 81: 10–19. Woods, C. 2006. Bilingualism, Scribal Learning, and the Death of Sumerian. In S. Sanders, Margins of Writing, Origins of Cultures, pp. 91–120. Chicago: The Oriental Institute of the University of Chicago.

OLD WINE IN NEW WINESKINS? HOW TO WRITE CLASSICAL EGYPTIAN RITUALS IN MORE MODERN WRITING SYSTEMS Joachim Quack Ritual Texts and How to Write them in Demotic Egyptian In Ancient Egypt, ritual texts are an important part of the culture. They are among the earliest attested longer texts, and they continue to be used down to the fijirst centuries CE. In quite a few cases, we can even see how the very same compositions were used for almost three millennia—or at least approximately the same compositions since in practice, some divergences in wording are the rule rather than the exception. This entails, however, that the language used in the ritual formulae is very diffferent from the language in actual use among the population—the efffect is likely to have been similar to the use of Latin in church service during medieval and modern times. Such a divergence has repercussions when it comes to the writing system used. There is no problem as long as you use traditional writing systems for rendering traditional formulae because the linguistic elements needed for classical Egyptian speech can easily be represented in hieroglyphic and hieratic writing. In actual realisation, hieratic ritual papyri even of the second century CE normally do not pose really serious obstacles to understanding them, even though the orthography of some selected few words might cause trouble. The situation becomes considerably more difffijicult once you go beyond this restricted circle. There was a growing tendency to write such ritual texts, even if they were linguistically very much in a traditional register of classical Egyptian language, in the new medium of demotic script. Quite a few manuscripts of this type known by now, especially an ostracon (oHor 18) with a ritual for purifijication and protection of the king and the Apisbull from Memphis (Quack in press), the funerary papyrus Louvre E 3452 (Smith 1979), an ostracon from Thebes with a complex liturgical hymn (Smith 1977; Smith 1999), a stela from Akhmim with a solar hymn (Vleeming 1990; Vleeming 2004), a late Ptolemaic or early Roman wooden tablet with a hymn to the uraeus-deity (Widmer 2004), an ostracon with a hymn

220

joachim quack

to the sistrum-play (Quack 2001), and a relatively substantial number of liturgical compositions on papyrus from Soknopaiou Nesos, mostly only available through preliminary reports (Hofffmann 2002; Hofffmann in press; Widmer 1998; Widmer 2005; Widmer 2007), plus some funerary compositions (Smith 2009a: 392; Smith 2009b; Smith 2010; Smith in press). These texts have, on fijirst look, one overarching similarity: they look strange and quite incomprehensible to somebody accustomed to demotic texts; and even if they were transcribed back into hieroglyphs, they would look equally strange to somebody accustomed to ritual texts in hieroglyphic and hieratic writing. However, already now it should be stressed that while those are the compositions which abound in the orthographies I am studying here, now and again we can encounter them also in other, mainly literary, texts,1 be it as single archaic lexems in an otherwise “normal” demotic texts or as chunks of archaic language in a text which is inhomogeneous in its language, like the Ritual for Entering the Chamber of Darkness.2 There are even some plausible cases of “unetymological” writing making use of word-groups in administrative and documentary texts (Lippert & Schentuleit 2010: 364–365). The orthography in question is normally called “unetymological” by modern scholars. Its basic point is that the demotic sign-combination for writing one word is used for the writing quite another, etymologically unconnected word. In order to illustrate the point, I would like to provide a number of actual cases: The old noun bw “place” is written as by “soul”.3 The older negation nn (no longer in use in contemporary texts) can be written as the group n⸗n “for us” (besides another option of writing it with two one-consonantal signs for n).4 The preposition of place, in older Egyptian m, but shifted in Demotic to n, can, besides a more normal orthography as m or n.m in these texts, also be rendered with the group for i҆rm “together with”.5 This sounds quite 1  Sometimes also in everyday compositions, if these, in giving a priestly title composed with the epithet of a god, have to come up with a way of writing it. 2  For this composition, also known by the less appropriate name “Book of Thot”, see the edition by Jasnow & Zauzich (2005); and the study and translation in Quack (2007a; 2007b). 3  E.g., pStrasbourg 3vs, x+7, 12.13.14; see for the writing J.F. Quack (2007a: 289) and Smith (2009a: 394 note 32). 4  See Smith (2005: 161 note c) with further references. 5  See Vleeming (2004: 628f. and 632 note y); there are also cases of i҆rm in the papyrus Insinger (coming from Akhmim as well as the stela in question) which I suspect to be writings of the preposition n/m, see Hofffmann & Quack (2007: 363 note h).

old wine in new wineskins?

221

strange when considering only the historical transcription of the group, but in reality, the preposition “together with” had a contemporary pronunciation about like *nm, to judge form Coptic forms like ⲛ̄ⲙ̄. The older Egyptian preposition m-ꜥ.w “through, by” is replaced in normal Late Egyptian and Demotic by the form m-ṭi. In ritual texts, when it is still used, it has an especially wide variety of possible writings. One possibility is to write it like m-ꜥw “in greatness”, another like mw “water” (Quack in press a), and fijinally, it can be written just with one-consonantal signs as mw (Smith 2005: 144). A group6 looking at fijirst sight like pꜣ wꜣč̣-tp “the one with the green head” is used several times for writing pꜣw.t-ṭpi҆.t “fijirst primeval time”.7 The old expression ꜥšꜣ-ḫpr.w “of numerous shapes” is written with a group looking like ꜥš “to call” followed by a sequence of one-consonantal signs giving ẖrb (Hofffmann 2001: 225). We should, of course, be careful not to confuse genuine “unetymological” writing with cases of textual variants where in a ritual formula known elsewhere a genuine change in formulation has occurred. A case where this can become relevant has been discussed by Mark Smith (Smith in press). The original text runs as rṭ kꜣ⸗k m-bꜣḥ⸗k rṭ kꜣ⸗k m-ḫt⸗k, “The foot of your ka is in front of you, the foot of your ka is behind you.” (Attested as such, e.g., in PT 18b (spell 25)). In a demotic papyrus in the Bodleian library transmitting a copy of this ritual formula, the noun rṭ “foot”, is written as if it were the verb rwč̣ “flourish”, giving the sentence a surface meaning “may your ka flourish before you, may your ka flourish behind you”. Since both words are current in demotic, and furthermore even have signifijicantly diffferent pronunciations (Coptic form of rṭ is ⲣⲁⲧ; Coptic form of rwč̣ is Sⲟⲩⲣⲟⲧ with a metathesis, the earlier form would be *ⲣⲟⲩⲟⲧ; other dialects have BMⲉⲣⲟⲩⲟⲧ),8 I fail to see why the one should be used as deliberate unetymological writing of the other. Rather this is a case of the classical phenomenon that Egyptian texts can have shifts in formulation which lead to the creation of a genuinely new sense (Pries 2011). The fact that in the case of this rather frequently attested text this would be the only manuscript to show such a deviation cannot

6  By the term “group” I designate a demotic line-sequence which goes back to a combination of several signs (normally two- or three-consonantal signs with their phonetic complements) but which, on a synchronic perspective, cannot be meaningfully broken down into its original constituents. 7  Widmer (2004: 668 and 678 with note 47); Quack (2007a: 281 with note 71); for a similar hieroglyphic writing see Quack (2009d: 141). 8  The letters S (for Saidic), B (for Bohairic) and M (for Middle Egyptian) indicate different Coptic dialects.

222

joachim quack

count as a counter-argument since all experience with textual criticism shows that many variants are products of a single scribe and attested only once. Necessary Distinctions of Different Cases of Writing Words in Demotic Egyptian In order not to confuse the issue, it is necessary to draw some clear distinctions which have not always been respected in previous discussions. Firstly, there are words which do not have a well-documented earlier attestation or fijixed orthography but only become normal in Demotic. Those have, of course, no established traditional orthography at all. The normal way is to write them simply with one-consonantal signs plus the appropriate determinative. In rare cases, a short demotic word-group with fijixed vocalic value can be incorporated, as in sy-thy.ṱ.t “danger” where the fijirst group is the one for sy “son”.9 Since such words do not have any traditional “etymological” form of orthography, we can obviously not speak of an unetymological writing at all (contra Stadler 2006: 202 n. 4). The second case are words which are attested in earlier Egyptian texts and also normal in demotic texts, but do not continue to use the same orthography. Those have not, up to now, been included in any discussion of “unetymological writing” in demotic, simply because demotists are too used to encounter them in their new orthography to feel in any way surprised by their outward form. The most frequent reason for coming up with a new orthography is that the traditional word-sign is no longer in use in demotic writing. This leads to applying a new orthography, normally one with one-consonantal signs based on the actual pronunciation. A god case is the word for “fijight”. In older Egyptian it occurs as ꜥḥꜣ with the word-sign . This sign does not exist in demotic. Due to regular phonetic developments, the contemporary pronunciation was ꜣḥ and this is how the word gets written in demotic orthography. In some cases, we can observe that the switch from a traditional to a modern writing is not coterminous with the rise of demotic writing as such, but only occurs in later stages. Especially early demotic writing,

9

 The word has no certain pre-demotic attestation; it might be recognized in the group-

writing

! ©ÒTihaTi‘¬

(stht) in pBrooklyn 47.218, 135, 2, 12.

old wine in new wineskins?

223

which is comparatively close to traditional hieratic forms, still makes use of some groups abandoned later. Thus, we still fijind the word bꜣ.w “divine

HÒ»'

in pRylands IX 24, 17; wrath” written very traditionally as 25, 5 with a group which did not continue as part of the demotic graphic inventory, while later texts at least from the Ptolemaic period onwards write the word with one-consonantal signs as bw(ꜣ).10 A subtype of this case are words where phonetic shifts of individual words have made it impractical to further use their traditional orthography even if that still exists for other derivations of the same root. A good example is the word ḫpr.w “shapes, forms”. Originally, it is derived as a noun formation from the verb ḫpr “to exist, to take shape”. Both the verb and the noun are still used in demotic (and Coptic) language, and there is a specifijic demotic group for the verb consistently used. However, for the noun a sound shift plus later a metathesis have occurred, leading to a writing with one-consonantal signs as ẖbr or later ẖrb, Coptic ϩⲣ̄ⲃ. The group for the verb is no longer used in the noun. We can also adduce the case of the word ẖry “what is down”. This is clearly a derivation of the preposition ẖr “under”. But while older Egyptian writings consistently show the connection by using the same core part of the writing, in demotic the word “what is down” is always written with one-consonantal signs, not with the historical group used in the preposition. The reason is probably that the preposition ẖr, dropping the fijinal r in pronunciation, became ϩⲁ in Coptic, while the adverb, pronounced ϩⲣⲁⲓ, sounded quite diffferent. In rare cases, especially when a phonetic development is not generalized in Egypt but confijined to certain regions, we also can have the situation that the historical form (with a fijixed group) is normal while an unetymological form corresponding to the sound-shift is very rare. We can produce the example of nb “lord” where the group is normal, while we rarely have a form np written with one-consonantal signs and corresponding to a shift to ⲛⲁⲡ attested in some Coptic upper Egyptian dialects.11 Perhaps the choice of the writing has also to do with the diffferent treatment of the word according to whether it is used in the status absolutus or the status constructus.

10

 For the correct etymological connection of the often misunderstood word, see Smith (1987: 119). 11  Attested pRhind I 5d6, see Quack (1991: 97, n. 30). The phonetic reason for this writing still goes unrecognized in Smith (2009a: 643 n. 98).

224

joachim quack

This leaves only the third type as really relevant, namely words no longer a normal part of active language during the Late, especially GraecoRoman period. Even with them, we have to be careful what is still used and what not, since that can involve quite similar-seeming words. A good case in point are nominal derivations of the ancient root wč̣ “to order, to decree”. While a masculine word wt “order” (ancient wč̣) is still functional in Demotic, its feminine counterpart (ancient wč̣.t “decree”) had died out. When the feminine word occurred in some religious compositions, the only option for writing it actually attested is using the word wt.t/wty.t “calamity” (ancient wṭ.t).12 Methodological Considerations After having singled out what is relevant and what not, we can now focus on the actual phenomenon that writings usual for one word, normally a common demotic one, are used for a quite diffferent, uncommon one. Detecting all relevant cases can be hampered by a methodological problem. In some cases, the new orthography does not produce any acceptable overall sense in the text, so that it is relatively easy to arrive at the conclusion that it must stand in an unetymological way for something quite diffferent. The only real problem is to fijind a good solution for what is actually meant. It can be much more difffijicult if the new writing does not, in itself, produce nonsense but might be considered as a semantically viable formulation. If no hieroglyphic or hieratic parallel with a clear orthography is at hand, the decision can become difffijicult. In such cases, the diffferent temper of individual scholars can lead them to varying approaches. Normally, they argue that the writing should be taken at face value “as it stands” (Smith 2009a: 278 n. 43). This sounds at fijirst like a solid way of handling a text. On closer inspection, however, it can turn out to be more problematic than it seems. If we say, in deciding how to semantically decode a specifijic writing, that we take a certain writing “as it stands”, this implies that the ancient writer has fijixed a default value for a certain word, namely a “normal”

12

 See Smith (1987: 117). Erroneously Stadler (2006: 202) who has not distinguished the masculine and feminine form. His rejection of the understanding of wty.t as “decree” in pVienna 12006 is ill-founded, as the “decree of Re” mentioned in that text is a composition attested also in hieroglyphic texts, see Quack (2008: 367 n. 221).

old wine in new wineskins?

225

orthography, and would only in exceptional cases resort to another, “unetymological” writing for it. This is, of course, not how things really stand. In all cases I have presented here for uncommon words being written by the use of groups for quite diffferent words, there is no demotic attestation at all for the use of the historical writing of the uncommon word. Thus, the scribe had no other choice for the uncommon word but to write it with the signs in use for the common one. This means that, if we have a text known to draw on uncommon vocabulary, in principle the one orthography can stand equally well for both words, and only considerations of context can provide arguments for a decision. Possible Explanations of the Writing An important question is of course why such a way of writing came into being. We can probably exclude the hypothesis that it was just an intellectual exercise where one priest tried to impress on another with his cleverness. Certainly the reading culture of these manuscripts would not be apt to such an attitude; after all, they were intended to be used in rituals with a clear goal for the community, not read for the fun of a specifijic individual. We can be a bit less certain that it was not, in a more serious variant of this, a trial item for checking the competence of priests. There is one piece of positive evidence that reading competence could be tested by the administration. In a Greek-language papyrus from Tebtunis (pTebtunis 291; Grenfell, Hunt & Goodspeed 1907: 54–58), we have a document concerning some adolescent candidates for priesthood. In most cases, the descent from a priestly family by the side of the father and the mother is the sufffijicient basis for allowing them to be circumcised, an operation at that time offfijicially forbidden and only allowed on application for Egyptian priests (for whom it was a necessary prerequisite for temple service). In one of the cases, however, the candidate does not prove his priestly descent but shows his competence by being able to read the hieratic and the Egyptian script from a hieratic book fetched from the temple.13 I suppose that the terminology means Hieratic and Demotic writing. This proof by competence is explicitly justifijied by reference to a recent imperial memorandum allowing this procedure alternatively to the proof by

13

 For the interpretation of this passage, see Quack (2005: 101).

226

joachim quack

descent. Thus, reading competence in rituals was of importance when it came to the question of being allowed to priesthood, and being a priest meant material benefijits. So there is some element of social competition for good jobs tied up with competence in reading ritual texts, and making the job more difffijicult would in efffect mean to diminish the fijield of potentially successful candidates. Still, the context of the manuscripts in question is not really in favour of such a setting. One line of thought applied to these orthographies was that they would act as a sort of internal commentary where writing was in itself giving an additional layer of meaning to the basic text (Smith 2009b: 357). To cite one proponent of such an approach: “Unetymological writings are capable of endowing a text with multiple levels of meaning. Moreover, they can refer to beliefs and concepts recorded in other texts and allow these to be brought into relation with it. I suggest that their particular contribution to ritual texts where the aim is to actualise both what is said and written, is to enhance their reifying power and extend the range of what they can actualise by bringing these other things into play. If Bodleian MS. Egypt. a. 3(P) is to be assigned to this category of texts where the manner in which a word is written is just as important as the manner in which it is pronounced, then it is not difffijicult to see why unetymological writings appear with such profusion in it.” (Smith in press). I am not quite convinced that this line of reasoning is entirely correct, or at least it does not seem to hit the mark for the primary relevance, and some other scholars share my doubts (Hofffmann in press). If unetymological writing was primarily a means of laying a commentary on the text itself and expanding its semantic relevance, we would expect it to occur in relatively equal measure all over the text, for all sorts of the vocabulary regardless of being archaic or still in common use, and especially for the semantically most loaded verbs and substantives. The real distribution is quite diffferent. The new and surprising orthographies are largely limited to those words for which there is no established demotic Egyptian written form, and this often goes concomitant with the fact that these words have died out in contemporary speech, or at least the specifijic forms of the words are no longer in use. But within this fijield, unetymological writings are by no means concentrated on words with highly loaded semantic content which would be in the centre of interpretation. They can also occur with prepositions or the negation or formative grammatical elements. Furthermore, there are too many orthographies which do not really give a satisfactory new level of meaning to the text. If, for example, the preposition “by” is written as “water” or “in greatness”, this hardly ever provides

old wine in new wineskins?

227

any acceptable new possibility of interpretation. Furthermore, while the cases where words with quite diffferent meaning are used have been more in the focus of the discussion, we should not overlook the fact that in most cases, the new writings are just plain sequences of one-consonantal signs without any deeper meaning at all. Another important aspect is that in the great majority of cases, we have only one or at most two diffferent “new” orthographies for the words in question,14 and those are used consistently, without consideration of the specifijic context. Obviously, such an application makes it relatively difffijicult to consider them as a commentary, as the essence of a commentary is to adapt to the specifijic context it is commenting. Actually, there is a remarkable stability over time and regions attested for many cases in question. We have, e.g., the ancient word č̣w “evil, bad”. In Demotic, this is written with the group for tꜣ “land”, optionally with the added determinative of the dying man. We have this writing in the following texts:15 – pBM 10507, late Ptolemaic or Early Roman, from Akhmim.16 – Tablet Louvre E 10382, early Roman period, perhaps from Dendara.17

14  A rare case of two diffferent writing possibilities is the word gmḥsw for a sacred falcon which can be written as gm-ḥs “fijinding praise” in the book of Thot and as gm-ḥs “fijinding excrements” in pVienna D 6951 (diffferent determinatives used for the second part), see Hofffmann in press. The reason for the divergence is perhaps that ḥs “praise” gradually fell out of normal use (it is hardly attested in Coptic). In the case of the four diffferent writings for the preposition m-ꜥ.w, local dialect diffferences are relevant, including the question which dialect had already lost the specifijic sound ꜥ. 15  In principle to be added to the documentation is an unpublished hieratic manuscript of the Book of the Temple (from Soknopaiou Nesos in the Fayum) where reversely č̣w is written in an expression which etymologically clearly contains tꜣ “land”. Also, a difffijicult hieratic text shows a variation between tꜣ and č̣w in diffferent manuscripts, see Goyon (1999: 91 n. 52 and pl. XXXIX [51, 7]); with the latter probably being the better reading. 16  Quack (1999a: 41). I would insist on this interpretation against the opinion of Smith (2009a: 262) who thinks that the text makes good sense “as it stands”. Firstly, while formulations like “you will be save from all evil things” are very frequent in Egyptian texts, an expression “spirits of the earth” is otherwise completely unattested (and “spirits on earth”, as he tries to translate now in Smith (2009a: 262), would be i҆ḫy ḥr pꜣ tꜣ), and secondly, the orthography as tꜣ is the only attested writing of č̣w “evil” in demotic texts. For writings of i҆ḫy “thing” as if it were i҆ḫy “spirit”, see, e.g., the expression swḥ- i҆ḫy heading the indications for the manual practice and the objects used in pMag. LL which are obviously to be understood as “collection of things” but are normally written as if they meant “collection of spirits”; for the translation see Quack (2008: 334 with note 17). 17  Widmer (2004: 664). The actual place of origin is unknown but the best parallels for the text are hieroglyphic inscriptions from Dendara. In any case, the writing seems upper Egyptian, not Fayumic.

228

joachim quack

– pCarlsberg 182, second century CE, from Tebtynis in the Fayum (Osing 1998: 211 and 212 note e). – pVienna D 6527, second century CE, from Soknopaiou Nesos in the Fayum (Quack 1999b: 461). – pMag. LL, third century CE, from Thebes (Grifffijith & Thompson 1904– 1909, volume III, p. 92). Tests of Frequency In order to put the writings into a correct perspective, some statistics about their relative frequency are needed. I have chosen to limit my sample to two single texts, namely the writing tablet Louvre E 10382 and the ostracon BM 50601, and that for several reasons. Firstly, the limited amount of data contained in them makes it faster to process them and to formulate preliminary hypotheses which can then be checked against the bulk of the documentation (which would be very difffijicult to analyze without any guiding premise). Furthermore, both have been treated in full in good recent publications, and most especially, the text contained in them is known from hieroglyphic parallels, thus there is rarely any doubt about the semantic interpretation of the actual writing.18 The writing tablet contains about 63 diffferent words (some occurring several times). Of these, about 32 (including personal sufffijixes) are words at home in demotic and written in a completely normal way; either with historical groups or as sequences of one-consonantal signs which are the accepted way of writing these words also in non-religious texts. 28 are written with one-consonantal signs,19 either totally or for a relevant part of them,20 and not part of the normal demotic vocabulary.21 For none of those words, there would have been an obvious historic demotic group at hand to write them. 18  These criteria would also apply, in theory, to the ostracon Corteggiani 1 and the stela from Akhmim, but those texts do not contain enough relevant orthographies to make them meaningful items for statistical analysis. 19  The use of m for the preposition has been included here, as the normal demotic orthography for this preposition would be n. The presufffijixal form i҆m⸗f has been counted separately. 20  I have counted the instances of h̭ꜣ and m-h̭ꜣ as phonetic alphabetical writing, not as unetymological writing proper, contrary to Widmer (2004: 677), since h̭ꜣ with the determinative of the man with the hand to the mouth does not correspond to any normal writing for h̭y “to be elevated, high”. 21  The demarcation between words forming part or not part of demotic vocabulary can be in doubt for some of them, but this is unlikely to substantially change the picture.

old wine in new wineskins?

229

Only three cases present words written, either as complete phonetic part or signifijicant part of it, with a group otherwise serving to write a clearly unrelated demotic word, namely the following: the word “evil, bad” (ancient č̣w) is written with the group for tꜣ “earth”, but followed by its appropriate determinative of the dying man. The word “plebs” (ancient rḫy.t) is written as ꜣl-ẖe.t, making use of the group for ẖe.t “body” for the fijinal part of the word. The word pꜣw.t ṭpi҆.t “fijirst primordial time” is written as ⟨pꜣ⟩-wt-ṭp “the one with the green head” (or, perhaps even more probably, “the fijirst papyrus plant”). All three writings are not specifijic to this single text but well attested in other manuscripts as the general and typical writing of the word in question whenever they occur in demotic. In the fijirst one, the use of the specifijic determinative distanced it so much from the word “earth” that clearly there was no intention to create an ambiguity or deeper meaning. The second one obviously does not have any really “deeper meaning” since the fijirst part ꜣl, written alphabetically and without determinative, does not mean anything at all. In the third one, a connection with Osiris as the “green-headed” god,22 proposed for other attestations of the expression (Smith 2005: 210), does not seem particularly relevant for the hymn in question where Osiris does not play any role. Thus, in this case, the image is crystal-clear: There is no predilection at all to use “meaningful” writings with word groups to create an additional layer of associations or commentary. Rather, you would simply use oneconsonantal signs to spell out the word in question (of course according to current pronunciation, not to the historical old form). If available, word-groups of fijixed vocalic value would be used in order also to indicate the vowels for the recitation, but those were few and far between them, and since all were the only actually attested demotic orthographies of the words in question, the use of this writing would not have evoked some esoteric deeper hints to a reader accustomed to this sort of texts (as most of their users would certainly have been). The ostracon presents a somewhat longer text. There are about 115 different words in it. Of those, about 65 (including personal sufffijixes) are written in a completely normal way. About 30 are somewhat alphabetically

22  As a matter of fact, although Osiris is iconographical depicted with a green head (for reasons discussed by Banaschak & Grothofff 2000), actual attestations of an epithet wꜣč̣-ṭp are not only enormously rare but even never refer to Osiris at all, see Leitz et alii (2002: 262).

230

joachim quack

written, although sometimes only for parts of the words.23 The number of writings which might be called “unetymological” comes at about 20.24 On fijirst sight, this looks like a signifijicantly larger number than in the writing tablet. However, several points have to be kept in mind. Firstly, some writings are bordering on cases which can hardly be called “unetymological” since, while at variance with traditional hieroglyphic spelling, they are much at home in demotic texts outside the religious sphere, e.g., i҆w (written like “income”) for i҆ꜣw “praise”.25 Furthermore, this spelling with complete words is by far the most frequent with words like particles and grammatical elements which should be semantically the least loaded. Thus, we have tw⸗t (like the proclitic pronoun of the 2 sg. fem.) for the conjunction č̣r,26 n-na (like “of those of ”) for the plural form of the genitive n.w, ⸗n like the sufffijix “we” for the formative element of the sč̣m.n⸗f-tense, nti҆ (like the relative converter) for the feminine form of the genitive n.t, w-wnw.t (like the word “hour” with an additional alphabetic w in front of it) for the auxiliary verb wnn “to exist” in a certain tense, i҆p (like the verb “to count”) for the copula of the nominal phrase,27 tw⸗t-s.t (like “you gave a place”) for the formation ti҆-si҆ “while she is”, mḥ (like he verb “to fijill”) for the compound preposition m-ḥꜣ.t “before”, n-n⸗n (like “for us”) for the negation, and tw-n⸗i҆ (like “gave me”) for the conjunction tnw “every time when”. Firstly, this leaves very few items where it could be even seriously discussed if the new writing gave an intentional layer of meaning. Secondly, the very fact that more than half of the cases where complete demotic words are used for unetymological writing come from this semantically rather “empty” category of words demonstrates clearly that considerations quite diffferent from an intention for “commentary” or “deeper meaning” were the driving factor. The following are the actual cases with nouns and full verbs: The participle of the verb nbi̯ “to fashion” is written with the group for nb “lord” (l. 4); the verb in question is not otherwise used in demotic. Here, an “added meaning” would in principle make sense as a creator-god could justly be called “lord”. 23

 This concerns sequences like s-šsp-p for sšp “light” and sh̭-ḫpr for sḫpr “produce”.  Assuming that the tw⸗n-n⸗f for nč̣.n⸗f is more likely to be a real variant than an unetymological writing. 25  For demotic attestations of this word, see Ryholt (1999: 39). 26  Hardly a very bold writing, since m-tw or even tw are attested demotic writings for old m-č̣r. 27  For other cases of this orthography, see Hofffmann (2001: 226). 24

old wine in new wineskins?

231

The word ꜥwy “limbs” is used for ꜥ.wi҆ “the two arms” (l. 4); the dual is no longer existent in demotic and could not have been expressed correctly by any writing of the word really meant. The semantic area of the word actually used for the writing is so close to the original one as to leave little leeway for an added meaning. The expression ḳꜣy.t ḳꜣi ̯.t “high hill” is written as gy n kꜣ “form of a bull” (l. 6). For the fijirst word of this expression, a writing as gy “form” is also attested in Mythus Leiden 10, 13. The second is otherwise unattested as adjective. The new writing, taken at face value, would have produced an acceptable sense in the context, and, were it not the known hieroglyphic parallels, might even have been accepted as what the text was intended to mean. The word ẖnm “well” has been written at the end of l. 6 and afterwards been erased. It does not correspond to anything in the attested hieroglyphic parallels but could reasonably stand for the verb ẖnm “to unite”. This verb is not part of the normal demotic vocabulary and all attestations of it are showing metathesis and sound-changes. Thus, there would have been no other orthography available. A literal translation as “the well of the ennead is in adoration for him” does not really make sense. šy.ṱ⸗f “his nose” is used for šw.ti҆⸗f “his feathers” (l. 9). Here also, the word is not used in normal demotic. In the context, where the illumination of the sky is treated, the writing would hardly give an appropriate additional layer of meaning. The expression mṭw rʾ “word of the mouth” is used for the verb “to be successful” of which the most ancient form seems to be mʿr but which in later periods was usually written as if it sounded like mʿrṭ (l. 10). This verb is otherwise unattested in demotic or Coptic. A “literal” reading as “great is the speech of the mouth” would be possible in the context. The city-name nw.t “Thebes” is used for the sky-goddess nw.t but with added determinatives of the sun-disk and the divine (l. 22). In this case, normal writings would have been at hand and the reason for the graphic choice is not quite obvious (it might have been the intention to indicate a diffferent pronunciation but we have no clear evidence for that). In any case, the determinatives added preclude any efffort at just taking the writing to actually mean “Thebes” although in a hymn to Amun, the main god of Thebes, the writing can be considered appropriate. Thus, while some of the actually attested unetymological writings could be analyzed to give an additional layer of meaning, such an approach cannot convincingly be upheld for all of them.

232

joachim quack Preliminary Conclusions

We can now resume the results from these two texts into some preliminary hypotheses and the consequences for interpretation they entail. Firstly, even in most of the liturgical texts in classical Egyptian, if demotic writing is employed, more than half of the words still appear in the regular demotic orthographies for those very words.28 In itself, that proves nothing more than the principal continuity of Egyptian lexems through times. The fact becomes more meaningful, however, if we turn it the other way round: if a normal demotic writing for the word was available, normally no demotic scribe bothered to make up any unetymological writing for it.29 This, in itself, is a serious indication that there was no dominating drive for achieving an additional layer of meaning in those cases where “unetymological” writings where employed, because if that had been the case, why would the scribes hardly ever have worked it on words for which a traditional orthography was available for them? The fact that sometimes similar writings were employed even in administrative and documentary texts where clarity would be the intention, not the creation of ambiguity and second meanings, also militates against the “commentary”hypothesis. Secondly, for words which did not have any traditional orthography in Demotic (and did not actually exist any longer in ordinary speech), the far more common solution was to employ purely alphabetical writings which did not allude to anything else at all. There was no urge felt by the scribes to look for writing solutions which could give additional layers of meaning. Thirdly, writings employing groups for completely unrelated other words were most frequent with semantically rather “empty” prepositions and grammatical elements. Fourthly, when they were used for nouns or verbs, in most cases there was no really convincing additional meaning 28  The only likely exceptions are texts containing mainly recondite divine and geographic names and epithets. 29  A rare exception is the divine name Ḥw.t-Ḥr “Hathor” which occurs once with the fijirst element written as ḥč̣ “silver” in Mythus Leiden 22, 24 and 25, as noted by Widmer (2004: 681). The writing provides a perfect indication of the phonetics if we suppose a state of the vowels as in the Saidic or Bohairic dialect where the word is pronounced ϩⲁⲧ (it would not function in Fayumic or Akhmimic where the form is ϩⲉⲧ), but semantically, it seems rather weak, since Hathor is the “golden”, not the “silver” goddess in Egyptian imagery. This writing is used only for Hathor linked geographically to Thebes and Dendara, and it might be a conscious efffort of the scribe to indicate a particular pronunciation at home in that area.

old wine in new wineskins?

233

which could be produced by the process. Finally, the only really constant factor of “unetymological” writings using complete demotic word-groups is phonetics. Wherever we have sufffijicient external evidence for the pronunciation of the words in question, we can conclude that the word used in writing really had the very same sound-sequence or a close approximation of it.30 We should also note one additional point: Classical-Egyptian language ritual texts in demotic script do have “unetymological” writing on a regular basis. But contemporary classical-Egyptian ritual manuscripts in hieratic script, sometimes even of the very same compositions, do not normally have them. It would be quite surprising if the urge to lay an additional meta-layer on a text were so much greater with demotic than with hieratic scribes. This is most evident in the case of papyrus Bodleian MS. Egypt. a. 3(P) which has some hieratic and some demotic ritual texts. Only the demotic ones are characterised by “unetymological” writing. Why would the scribe have refrained from using it for the hieratic parts of the manuscript if he had really a substantial interest in providing such an additional layer of meaning? However, the fact of the choice between hieratic and demotic writing to use for the ritual manuscripts is meaningful in itself. If some scribes preferred demotic script for ritual texts even though it forced them to use complicated non-etymological writings, there must have been other advantages to be gained (Hofffman in press). One point is perhaps the better legibility of demotic script. This might seem surprising to us, since modern Egyptologists normally deem demotic to be far more difffijicult. But the ancient training seems to have been diffferent, with demotic as the normal writing system taught during the Late Period. Even more important is probably another point. The crucial part of a ritual is to recite it with correct pronunciation. Understanding the semantics of the spells by the recitator was much less an issue. This means that demotic script, by the very fact of being unetymological, was helpful for the performer. Instead of having to memorise lots of words not part of his active language, he could just use a manuscript with relatively clear indications for correct pronunciation.

30  Due to constraints of space as well as the highly special discussion needed for it, I refrain from demonstrating this in more detail; already Hofffmann (2001) has provided several good examples.

234

joachim quack Additional Evidence

The question of competence in a somewhat dead earlier phase of Egyptian language cries for introducing another piece of evidence up to now not much drawn into the debate about the sense of demotic “unetymological” writing. This is an onomasticon from the second century CE which lists of words arranged according to semantic groups (Osing 1998). The basic part of the text is written in hieratic, with a relatively classical orthography. But above many of the words, there is a supralinear gloss, often in demotic, and sometimes also in Greek letters with some additional demotic oneconsonantal signs for sounds not present in Greek (Osing 1998: 40–50). In most cases the glossing is concerned with questions of pronunciation of the whole word or parts of it (especially the stressed syllable which would be the only one to contain a full, unreduced vowel). In other cases, there can be also an indication of meaning, realised by the label “sort of . . .”. The demotic forms for the pronunciation are obviously to be seen on a level similar to the writings in the ritual texts we have discussed so far. Some examples should show this: For the verb ḥri ̯ “to prepare” we have as a gloss the divine name of Horus, Egyptian Ḥr. The word ꜥḥꜥy.t “noon time” is glossed by i҆ḥ.t “cow”. For mṭw “stafff ”, we have the sign for the number 10 (mṭw). A verb prs of uncertain meaning is glossed by pꜣ rsi҆ “the south”. ḫmy “enemies” is glossed as šmw “parents in law”. The obvious reason for these glosses is the training of ritual specialists. With its collection of words, many of them no longer in current use, and their classifijication according to semantic categories, this is a classical case of a reference tool. Even more, the second half of the papyrus collects specifijically fundamental religious knowledge. This clearly shows how it was a manual for the specialist and helps further to situate the glossing. The knowledge of meaning and pronunciation of classical Egyptian was a precarious thing and in need not only of training as such, but even of written fijixation. One point is now of crucial importance: in this text we have a collection normally of individual words or at most very short statements. Thus, the idea proposed for the ritual manuscripts, that the “unetymological” writing should give an additional layer of meaning to the text, is hardly pertinent. After all, here we have single words which could be used in quite diffferent contexts, and they do not form a coherent text. How could a single “unetymological” writing for any one of them efffectively produce

old wine in new wineskins?

235

a context-sensitive commentary or additional meaning? Thus, only the advantage of indicating the correct sound more clearly comes out as overarching reason for choosing the demotic writing for rituals in classical Egyptian script, including either purely one-consonantal indications of the actual pronunciation, or groups for quite diffferent words of identical or closely similar sound. Theoretical Questions Thus, we can certainly arrive at the conclusion that oral performance played an important part in choosing this notation. However, if we regard the situation from a theoretical point of view, we could say more about what happens here. The point is that we have become accustomed to regard certain Egyptian orthographies as “regular” because they got relatively fijixed during the formative early periods of Egyptian writing. If we look at it from the synchronic point of view of the early Egyptians, for most words, in the beginning, writing was in some way “unetymological” because the basic signs used were gained by abstracting sounds and sound-sequences from the concrete objects depicted and applying them to these sound-sequences in general, independently of etymological connection. Establishing any sort of orthography meant creating writings which could become future norms but were not, at the time of their fijirst establishment, already canonical—they only became the norm because time proved their practical usability. What we have now with the writing system used for classical Egyptian texts in a more modern writing system is, I think, not fundamentally different from what went on during the formative phase of Egyptian hieroglyphic writing. Where applicable, signs with the value of several sounds were used; else recurrence to simple one-consonantal signs was the option. The only serious diffference is about the basic material to be used. In the phonetic part of hieroglyphic writing, signs with the values of one, two or three consonants or whole words were available. By contrast, in demotic, for the signs with phonetic value, it only makes sense to diffferentiate between one-consonantal signs and groups (Quack 2009a; Quack in press b). Groups, however, practically always represent a writing of a complete word. Thus any writing of a demotic word which is not confijined to oneconsonantal signs cannot avoid creating associations with a word. If the word is etymologically closely related, we would normally consider this as an etymological writing and not pay more attention to it. If, however, the

236

joachim quack

word seems substantially diffferent in meaning or etymology, we speak of unetymological writing and try to come up with explanations for it. Perhaps this is not the best possible approach. Shouldn’t we rather loosen our fijixation on the seemingly word-fijixed value of the demotic groups and go a step towards recognising them as signs for a sequence of several sounds, partially with fijixed vocalic value? After all, in those cases which happen too regularly to be explicitly noted and commented, we actually do practice such an approach. For example, the complete group for the word mn “to remain”, including the determinative, is often used in Roman period manuscripts as part of the writing of the word mny “daily”. The groups for the verbs i҆ni ̯ “to fetch” and i҆ri ̯ “to do” are used freely for indication the sounds of n and r in specifijic phonetic situations (Zauzich 1998: 748f.). Nobody has up to now considered such writings under the same angle as that of the “unetymological” writings, perhaps because they seemed so normal. But frequency of attestation and theoretical classifijication are obviously diffferent questions. Thus, the phenomenon of writing in demotic ritual texts looses much of its seemingly specifijic nature. It is bound up with a general restructuring of the writing system occurring with the onset of Demotic in which, once again, clocks were set back to zero, and for establishing orthography you would start with what signs for single consonants or sequences of sound you had, basing yourself on the actual pronunciation. The only serious difference to the fijirst formative period of Egyptian writing during the protoand early-dynastic period is that this time, vocalic value at least of some groups seems to have been fijixed while the early development was characterised by the enormous abstraction achievement of disregarding vowels.31 This might bring up another point which I have withheld up to now. I have only spoken of rendering Egyptian-language ritual texts in demotic writing. But there are also cases where demotic writing is used for Semitic ritual texts, most especially the famous papyrus Amherst 63 (Quack 2009b). If we look closely, on a structural basis things are not so diffferent, with the only exception that here of course there is no traditional Demotic writing using word-groups possible for the Semitic words at all. But also here we have the situation that most words are written simply with

31  As I have noted in Quack (2006), contrary to many theories of Graecocentric scholars, disregarding the vowels in writing was not simply a shortcoming but had real advantages, and besides it was an abstraction only possible if you clearly understood the diffference between consonants and vowels.

old wine in new wineskins?

237

one-consonantal signs while in statistically much rarer cases, complete Egyptian groups were used, and perhaps also there with a fijixed vocalic value. But the content as well as the language of the basic text obviously excludes that those Egyptian words were chosen to give an added layer of meaning. Coming to a conclusion about the whole question of supposedly “unetymological” writing, it seems that it was tackled the wrong way in scholarly discussion, as far as theoretical implications go. It is no more than the logical result of the structure of the demotic writing system which gave you only one-consonantal signs and complete groups. If you had to write a word without established orthography, you could either simply take one-consonantal signs or, if it fijitted in, a group. Using a group was much less frequent for the simple reason that it happened only infrequently that an etymologically completely unrelated word would provide a phonetically good fijit for the word, or part of it, you wanted to write. It would be advisable to drop the term of “unetymological” writing altogether and label these writings just as “phonological”. The theoretical background to the orthography was identical in all sorts of texts. The only reason why “unetymological” writing was so much more frequent in ritual texts than in other genres is that the traditional ritual texts contain many more words having died out in contemporary speech and being in need of creating an orthography for them. This means that there is no mysterious deeper meaning to the orthography of these ritual manuscripts. The only thing which remains is our technical difffijiculty to decode them, which is due to the fact that we are still at a pioneer level where we have to get better acquainted with the specifijic writings for many archaic words and formulations. Going Further On Going beyond the demotic writing for ritual texts with its many “unetymological”, or better “phonological” spellings, even a further step is possible, namely to apply the Greek alphabet to such a ritual text. After all, many scholars, especially those coming from classical studies, think that the Greek alphabet is quite superior to earlier writing systems, not least in its capability to render phonetically accurate and unambiguously the sound of speech (e.g., Havelock 1982; Havelock 1986; Powell 1991). The logic followed so far would indicate that using the Greek alphabet is a logical next step to use on Egyptian ritual texts since its use would be even clearer and

238

joachim quack

simpler for guarding the pronunciation of the spells than what could be expressed in demotic writing. As a matter of fact, there actually are a few compositions which turn this approach into a real practice. However, we have of necessity to add one element to the Greek alphabet to make it viable as a usable medium of rendering Egyptian language. There are some sounds not existing in Greek but important as phonemes in Egyptian. These have to be indicated in any text which seriously intends to notate Egyptian languages in a way enabling easy and correct pronunciation. Thus, the Greek letters are supplemented by a group of signs taken over from one-consonantal signs of the demotic script for the sounds in question. Especially in the earliest texts making use of such signs, you can still feel the groping for the best solution in this domain, with several possible variants of which only one is fijinally generalized in what becomes the normal Coptic writing system. These early and still experimental texts are nowadays known as “Old-Coptic”. However, I should stress that most of them are not directly relevant for my topic because as for the content, they do not reproduce age-old rituals but contemporary texts in a language closely mirroring contemporary speech. There is only one substantial text which really contributes to the question of traditional ritual texts, even if it poses enormous problems. That the text preserved in papyrus British Museum 10808. By now, there have been some articles and two monographs devoted to it, but we are far from having reached even a basic consensus as to what that papyrus actually is about.32 Most often, it has been understood as a sequence of three spells against fever-demons. I personally disagree profoundly and would like to interpret the text as a ritual for gaining favour and afffection. Such a divergence of opinion clearly illustrates that the papyrus is far from easily understandable on a semantic level, even if the column is preserved fairly completely and thus neither the legibility of the signs nor lacunae are a major source of problems. Perhaps I should say more specifijically that what has caused such serious problems is actually one column of the text which contains exclusively incantation formulae to be recited. There are some remnants of a second column, and while they are too damaged to yield much connected

32  Principle studies are Crum (1941); Volten (1953); Osing (1976) with critical reviews by Vergote (1977) and Shisha-Halevy (1980); Sederholm (2006) with critical reviews by Depuydt (2008) and Quack (2009c).

old wine in new wineskins?

239

sense, they are in themselves quite clear and understandable, and this is due to one basic trait: They are written in demotic Egyptian script and employ contemporary language. This fact can by now be augmented by further evidence. There is a still unpublished fragment in the papyrus collection of the EES at Oxford (105/124) which represents almost certainly the upper half of the second column, quite well preserved. In it, the title, the application and the manual instructions are written out in demotic, only the actual incantation makes use of Greek letters, in this case without any obvious background in Egyptian language. This gives us one important key for the overall interpretation. The scribe and user of the papyrus himself was capable of writing and reading demotic but did not deem it the best option for writing down the spells. In this case we can be absolutely sure that he did not intend to convey an additional layer of meaning because a text written almost exclusively in Greek letters and demotic one-consonantal-signs could not possibly function as a commentary in the way alleged for the demotic ritual texts. Sederholm proposed in his recent edition that the text was alphabetic, but not vocalic and would not have been understandable for another scribe (Sederholm 2006: XIV). I tend to seriously disagree. The very aim of this text was to render as accurately as possible the real contemporary pronunciation of the spell in order to assure the most correct recitation. The use of vowels, in cases where the sense is well enough understood to make a check possible, seems correct and consistent, thus we can suppose that the document was actually intended to convey the sound of the recitation as closely as it was possible with the writing tools available. The break with traditional Egyptian writing traditions this entailed, however, clearly produces a shift in priority. While the phonetic realisation becomes even easier than with any sort of Egyptian writing, the semantic side of the text looses seriously. We can still suppose that an Egyptian ritual specialist would have fewer difffijiculties than a modern scholar, being more at home with such texts than we can ever hope for. But still the fact that the text is so difffijicult for us to decode is some indicator towards the problems it would present even to an ancient Egyptian. In this case, the difffijiculties are serious even though the use of additional signs derived from demotic one-consonantal signs meant that all existing sounds of the Egyptian language could be appropriately represented. We can imagine how much more difffijicult things become once we use only the existing Greek letters without additional indication of sounds unknown to Greek. This, however, is what would happen in magical manuscripts of the third to fijifth century BCE from Egypt. They are basically

240

joachim quack

written in Greek, especially for the manual indications to be performed. The recitations, however, could include sections noted in Greek writing but not simply being Greek. Rather, there was an admixture of Egyptian or Semitic parts. Obviously, semantically decoding these incantations is highly difffijicult, and modern scholars are far from understanding all of it, even the proposed etymologies for certain sound-sequences (assembled in Brashear 1995) are often controversial and rarely really convincing. Regardless of the cleverness of modern scholars in decoding them, however, what matters at least equally is how ancient practitioners and listeners reacted to them. We can suppose that quite soon after the forging of the incantations they had no longer any clear idea of what was semantically meant. We can even bolster this claim by recurrence to a well-known ancient discursive passage in Greek. It is found in a treatise where the ritual technique of theurgy is defended, which claims to be based on methods used by Egyptian priests for achieving direct contact with the deity. One point raised by critics was the sequences of “barbaric” words without any discernible clear meaning. The author of the treatise defends their use ( Jamblichus, De mysteriis VII 4f.). As principal argument he adduces that those sound-sequences are dear to the gods regardless of if they are understood by men (des Places 1966: 191–195). They are simply the fijirst ones used in contact with the deity, and their immutability was a virtue in itself. This probably tells us quite a bit about contemporary reactions in the third and fourth century CE, and it is the logical conclusion of the developments in the writing of Egyptian ritual manuscripts I have presented here.33 References Banaschak, S. & Th. Grothofff 2000. Osiris—der grüne Totengott, Göttinger Miszellen 175: 17–21 Brashear, W. 1995. The Greek Magical Papyri: An Introduction and Survey; Annotated Bibliography (1928–1994). In W. Haase (ed.), Aufstieg und Niedergang der römischen Welt, Teil II: Principat, Band 18/5, pp. 3380–3684. Berlin/New York: de Gruyter. Crum, W. 1941. An Egyptian Text in Greek Characters. Journal of Egyptian Archaeology 28: 20–31. Depuydt, L. 2008. Review of Sederholm 2006. Lingua Aegyptia 16: 349–354.

33  For the connection of Jamblichus’ statement with Egyptian realities, see already Quack (2004: 442).

old wine in new wineskins?

241

Goyon, J.C. 1999. Le papyrus d’Imouthès fijils de Psintaês au Metropolitan Museum of Art de New-York (Papyrus MMA 35.9.21). New York: The Metropolitan Museum of Art. Grenfell, B.A. S. Hunt & E.J. Goodspeed 1907. The Tebtunis Papyri, Part II. London: Henry Frowde. Grifffijith, F.Ll. & H. Thompson 1904–1909. The Demotic Magical Papyrus of London and Leiden. London: H. Grevel & Co. Havelock, E.A. 1982. The Literate Revolution in Greece and its Cultural Consequences. Princeton: Princeton University Press. —— 1986. The Muse learns to write. Reflections on Orality and Literacy from Antiquity to Present. New Haven/London: Yale University Press. —— 2002. Die Hymnensammlung des P. Wien D6951. In K. Ryholt (ed.), Acts of the Seventh International Conference of Demotic Studies Copenhagen, 23–27 August 1999, CNI Publications 27, pp. 219–228. Copenhagen: Museum Tusculanum. Hofffmann, F. in press. Der demotische Papyrus Wien D 6951. In J.F. Quack (ed.), Ägyptische Rituale der griechisch-römischen Zeit. Tübingen: Mohr Siebeck. Hofffmann, F. & J.F. Quack 2007. Anthologie der demotischen Literatur. Berlin: Lit-Verlag. Jasnow, R. & K.-Th. Zauzich 2005. The Ancient Egyptian Book of Thot. A Demotic discourse on Knowledge and Pendant to the Classical Hermetica. Wiesbaden: Harrassowitz. Leitz, Chr. et alii 2002. Lexikon der ägyptischen Götter und Götterbezeichnungen, Band II: o-b. Orientalia Lovaniensia Analecta 111. Leuven, Paris, Dudley MA: Peeters. Lippert, S.L. & M. Schentuleit 2010. Stoetis in geheimer Mission. Der Brief pBerlin P 8092. In H. Knuf, Chr. Leitz, D. von Recklinghausen (eds.), Honi soit qui mal y pense. Studien zum pharaonischen, griechisch-römischen und spätantiken Ägypten zu Ehren von HeinzJosef Thissen, OLA 194, pp. 357–381, pl. 72–74. Leuven, Paris, Walpole, MA: Petters. Osing, J. 1976. Der spätägyptische Papyrus BM 10808, Ägyptologische Abhandlungen 33. Wiesbaden: Harrassowitz. —— 1998. The Carlsberg Papyri 2. Hieratische Papyri aus Tebtunis I, CNI Publications 17. Copenhagen: Museum Tusculanum. des Places, É. 1966. Jamblique, Les mystères d’Égypte. Paris: Les Belles Lettres; 42003. Powell, B. 1991. Homer and the Origin of the Greek Alphabet. Cambridge, New York: Cambridge University Press. Pries, A. 2011. Die Stundenwachen-Riten im Osiriskult. Eine Studie zur Tradition und späten Rezeption von Ritualen im Alten Ägypten, Studien zur spätägyptischen Religion 2. Wiesbaden: Harrassowitz. Quack, J.F. 1991. Über die mit ʿnḫ gebildeten Eigennamen und die Vokalisation einiger Verbalformen. Göttinger Miszellen 123: 91–100. —— 1999a. Weitere Korrekturvorschläge, vorwiegend zu demotischen literarischen Texten. Enchoria 25: 39–47. —— 1999b. Review of W. Westendorf, Handbuch der altägyptischen Medizin. Orientalistische Literaturzeitung 96: 455–462. —— 2001. Ein Standardhymnus zum Sistrumspiel auf einem demotischen Ostrakon (Ostrakon Corteggiani D 1). Enchoria 27: 101–119, pl. 4. —— 2004. Griechische und andere Dämonen in den demotischen magischen Texten. In Th. Schneider (ed.), Das Ägyptische und die Sprachen Vorderasiens, Nordafrikas und der Ägäis. Akten des Basler Kolloquiums zum ägyptisch-nichtsemitischen Sprachkontakt Basel 9.–11. Juli 2003, Alter Orient und Altes Testament 310, pp. 427–507. Münster: UgaritVerlag. —— 2005. Ämtererblichkeit und Abstammungsvorschriften bei Priestern nach dem Buch vom Tempel. In M. Fitzenreiter (ed.), Genealogie—Realität und Fiktion von Identität, IBAES V, pp. 97–102. London: Golden House Publications. —— 2006. Die Rolle der Hieroglyphenschrift in der Theorie vom griechischen Vokalalphabet. In W. Ernst, F. Kittler (ed.), Die Geburt des Vokalalphabets aus dem Geist der Poesie. Schrift, Zahl und Ton im Medienverbund, pp. 75–98. München: Wilhelm Fink.

242

joachim quack

—— 2007a. Die Initiation zum Schreiberberuf im Alten Ägypten. Studien zur Altägyptischen Kultur 36: 249–295. —— 2007b. Ein ägyptischer Dialog über die Schreibkunst und das arkane Wissen. Archiv für Religionsgeschichte 9: 259–294. —— 2008. Demotische magische und divinatorische Texte. In B. Janowski, G. Wilhelm (ed.), Texte aus der Umwelt des Alten Testaments, Neue Folge Band 4. Omina, Orakel, Rituale und Beschwörungen, pp. 331–385. Gütersloh: Gütersloher Verlagshaus. —— 2009a. Difffijicult Hieroglyphs and unreadable Demotic? How the Ancient Egyptians dealt with the Complexity of their Script. In A. de Voogt & I. Finkel (eds.), The Idea of Writing. Play and Complexity, pp. 235–251. Leiden: Brill. —— 2009b. Egyptian Writing for non-Egyptian Languages and vice-versa—a short overview. In A. de Voogt & I. Finkel (eds.), The Idea of Writing. Play and Complexity, pp. 317–325. Leiden: Brill,. —— 2009c. Review of Sederholm 2006. Orientalistische Literaturzeitung 104: 27–33. —— 2009d. Review of D. Kurth, Einführung ins Ptolemäische, Teil 1. Die Welt des Orients 39: 140–149. —— in press a. Eine Götterinvokation mit Fürbitte für Pharao und den Apisstier (Ostrakon Hor 18). In J.F. Quack (ed.), Ägyptische Rituale der griechisch-römischen Zeit. Tübingen: Mohr Siebeck. —— in press b. Bemerkungen zur Struktur der demotischen Schrift und zur Umschrift des Demotischen. In M. Depauw (ed.), Acts Demotistentagung Leuven 2008. Ryholt, K. 1999. The Story of Petese, Son of Petetum and Seventy Other Good and Bad Stories, CNI Publications 23. Copenhagen: Museum Tusculanum. Sederholm, V.H. 2006. Papyrus British Museum 10808 and its Cultural and Religious Setting. Probleme der Ägyptologie 24. Leiden: Brill. Shisha-Halevy, A. 1980. Review of Osing 1976. Journal of Egyptian Archaeology 66: 181–186. Smith, M. 1977. A New Version of a Well-Known Egyptian Hymn. Enchoria 7: 115–149, pl. 18. —— 1979. The Demotic Mortuary Papyrus Louvre E. 3452. Dissertation University of Chicago. —— 1987. Catalogue of Demotic Papyri in the British Museum III. The Mortuary Texts of Papyrus BM 10507. London: British Museum. —— 1999. O. Hess = O. Naville = O BM 50601: An Elusive Text Relocated. In E. Teeters & J.A. Larson (eds.), Gold of Praise, Studies on Ancient Egypt in Honor of Edward F. Wente, Studies in Ancient Oriental Civilisations 58, pp. 397–404. Chicago: University of Chicago Press. —— 2005. Papyrus Harkness (MMA 31.9.7). Oxford: Grifffijith Institute. —— 2009a. Traversing Eternity. Texts for the Afterlife from Ptolemaic and Roman Egypt, p. 392. Oxford: Oxford University Press. —— 2009b. New Extracts from the Book of the Dead in Demotic. In D. Devauchelle & G. Widmer (eds.), Actes du IXe congrès international des études démotiques Paris, 31 août– 3 septembre 2005, Bibliothèque d’Études 147, pp. 347–359. Cairo. —— 2010. A Divine Decree for the Deceased (O. Strasbourg D. 132 + 133 + 134). In H. Knuf & Chr. Leitz, D. von Recklinghausen (eds.), Honi soit qui mal y pense. Studien zum pharaonischen, griechisch-römischen und spätantiken Ägypten zu Ehren von Heinz-Josef Thissen, OLA 194, 439–4454, pl. 81–83. Leuven, Paris, Walpole, MA: Petters, —— in press. Bodl. MS. Egypt. a. 3(P) and the Interface Between Temple Cult and Cult of the Dead. In J.F. Quack (ed.), Ägyptische Rituale der griechisch-römischen Zeit. Stadler, M. 2005. Das Ritual, den Tempel des Sobek, des Herrn von Pai, zu betreten: ein Ritualtext aus dem römischen Fayum. In H. Roeder & B. Dücker (eds.), Text und Ritual. Kulturwissenschaftliche Essays und Analysen von Sesostris bis Dada, pp. 149–163. Heidelberg: Synchron. —— 2006. Isis würfelt nicht. Studi di Egittologia e di papirologia 2: 187–203.

old wine in new wineskins?

243

—— 2007. Zwischen Philologie und Archäologie: Das tägliche Ritual des Tempels in Soknopaiou Nesos. In M. Capasso (ed.), New Archaeological and Papyrological Researches on the Fayyum. Proceedings of the International Meeting of Egyptology and Papyrology Lecce, June 8th–10th 2005, Papyrologica Lupiensia 14, pp. 283–302. Lecce: Congedo. Vergote, J. 1977. Review of Osing 1976. Bibliotheca Orientalis 34: 135–139. Vleeming, S. 1990. Translitterating Old Egyptian in Demotic. Göttinger Miszellen 117/118: 219–223. —— 2004. A Hieroglyphic-Demotic Stela from Akhmim. In F. Hofffmann & H.J. Thissen (eds.), Res severa verum gaudium. Festschrift für Karl-Theodor Zauzich zum 65. Geburtstag am 8. Juni 2004, Studia Demotica 6, pp. 623–637. Leuven, Paris, Dudley MA: Peeters. Volten, A. 1953. An Egyptian Text in Greek Characters. In Studia Orientalia Ioanni Pedersen sepuagenario a.d. VII Id. Nov. anno MCMLIII a colleges discipulis amicis dedicata, pp. 364–376. Kopenhagen: Munksgaard. Widmer, G. 1998. Un papyrus démotique religieux du Fayoum. P. Berlin 6750. Bulletin de la Société d’Égyptologie de Genève 22: 83–91. —— 2004. Une invocation à la déesse (tablette démotique Louvre E 10382). In F. Hofffmann & H.J. Thissen (eds.), Res severa verum gaudium. Festschrift für Karl-Theodor Zauzich zum 65. Geburtstag am 8. Juni 2004, Studia Demotica 6, pp. 651–686, pl. 51. Leuven, Paris, Dudley MA: Peeters. —— 2005. On Egyptian Religion at Sokonopaiou Nesos in the Roman Period (P. Berlin 6750). In S. Lippert & M. Schentuleit (eds.), Tebtynis und Soknopaiou Nesos. Leben im römerzeitlichen Fajum, pp. 171–184. Wiesbaden: Harrassowitz. —— 2007. Sobek who arises in the primeavel Ocean (PBM EA 76638 and PStrasbourg dem. 31). In M. Capasso & P. Davoli (eds.), New Archaeological and Papyrological Reearches on the Fayyum. Proceedings of the International Meeting of Egyptology and Papyrology Lecce, June 8th–10th 2005, Papyrologica Lupiensia 14, pp. 345–354. Lecce: Congedo. Zauzich, K.-Th. 1998. Ein vieldiskutiertes Wort im Titel des Hieros Polos der Königin Kleopatra III. In W. Clarysse, A. Schoors & H. Willems (eds.), Egyptian Religion, The Last Thousand Years. Studies Dedicated to the Memory of Jan Quaegebeur, Orientalia Lovaniensia Analecta 84, pp. 747–750. Leuven: Peeters.

SUBJECT INDEX abbreviation, 19, 196, 217 abecedary, 14–22, 27, 29, 47, 49, 50, 51, 71, 167 abgad, 13–16, 18–20, 24, 27–31, 33, 35–41, 43–47 abugida, see also alpha-syllabary, 7, 8, 23, 136 acrostic, 16, 19, 20, 51 adaptation, adapted, adapt, 6, 8, 9, 16, 18, 25, 26, 29, 33, 41, 47, 54, 55, 62, 83–101, 149, 161–164, 170, 227 administration, administrative, 1, 2, 7, 17, 44, 55, 120, 159, 163, 164, 181, 184, 201, 220, 225, 232 adoption, adopted, 15, 25, 30, 36, 65, 84, 103, 105, 106, 122, 123, 128, 150, 160, 172 aesthetic, 125 Akhmim, 219, 220, 227, 228, 243 algorithm, 70, 74, 77 allophone, 89, 96, 148, 194 alphabet, 1–54, 70–83, 85, 88, 101, 104, 105, 111, 114, 120–124, 127, 133–138, 141–148, 154, 162–164, 185, 218, 228–230, 232, 237–239 alpha-syllabary, see also abugida, 4, 5, 7–10 ambiguity, (un)ambiguous, 17, 25, 26, 29, 34, 84, 151, 160, 229, 232, 237 animal, 177, 195, 206–208, 210 anthroponym, personal name, 38, 176, 206 arapacana, 137, 138, 141–147, 151, 153, 162, 163, 165, 167 archaic, 12, 49, 220, 226, 237 assimilation, assimilated, 26, 86, 100, 101, 131, 136, 156, 186, 192, 193 bamboo, 6, 9 bilingual, see also multilingual, 83, 155–158, 201, 207, 214, 215, 218 bird, see flight biscriptual, see also multiscriptual, 154–156, 158, 161, 162 blue, 64 blunder, 91 book, textbook, 6, 77, 83, 84, 97, 100, 101, 103, 123, 150, 215, 220, 225, 227 bookkeeper, bookkeeping, 44 borrowing, borrowed, 1–9, 30, 36, 54, 69, 71, 72, 76, 86, 88, 99, 114, 122, 123, 128 bottom-up, 6, 108

bronze, 18, 29, 50, 156, 198 Late Bronze Age, 33, 37 buksi, see also vehicle, 57, 62, 63 buntai, 169 bureaucratic, 164 calendar, 42 calligraphy, calligrapher, 11, 32, 33, 50, 62 car, cart, see also vehicle, 57, 61, 62 cat, see Rongorongo circle, 12, 85, 87–89, 94, 96, 99, 102, 105, 108, 156 clay, 17, 18, 181 cluster of consonants, 25, 89, 94, 99, 108, 124, 136, 146–148, 153, 154, 159, 160, 186, 205, 212 coin, 139, 155, 156, 165, 167 colour, colouring, 191, 195–197, 201 compound, 187, 190, 194, 204 syllabic compound writing, 105, 106, 108, 109, 111, 120–122, 124, 126–128 copper, 165 cow, 208, 234 crossword puzzle, see games cursive, 4, 8, 31, 54, 108, 114–116, 118, 119, 122, 128, 153 decipherment, see also Rongorongo, 33, 53 default, 83, 88, 94, 121, 224 determinative, 222, 227–229, 231, 236 dextrograde, see also left-to-right, 16–18, 21 diacritic, diacritical, 2, 8, 12, 13, 85, 88, 94, 96–100, 116, 119, 121, 136, 146, 147, 153, 160–162 digraph, 12, 74 diphthong, 25, 84, 91, 107, 111, 135, 136, 152, 196, 203 divider, 21 dot, dotted, 12, 91, 94, 98, 99, 107, 108, 110, 119, 160, 161 ductus, 62 ejective, postglottal, 185, 194, 217 emperor, 177 empire, 154, 164 epigraph, 132, 133, 149, 150, 156–158 epigraphy, epigraphic, 24, 28, 36, 149, 156, 161, 164

246

subject index

epithet, 220, 229, 232 error, erroneous, 59, 224 experimental, 238 flap, 96 flight bird, 24 klm, 69 games crossword puzzle, 128 die, dice, 156 Scrabble, 128 geminate, doubling, 86, 114, 119, 136, 147, 153, 156 glottalized, 86, 215, 216 glottal stop, 21, 59, 71, 79, 90, 102 God, 57, 178, 195, 206, 207, 220, 229–231 name of a god, theonym, 176 goddess of scribal art, 207 gold, golden, 39, 232 government, 55, 57, 103 graecocentrism, graecocentric, 24, 26, 236 grafffijiti, 8, 31, 53–67 grammatogeny, 2, 4, 7, 105, 127 grammatology, grammatologist, 22–24, 27, 30, 33, 48 grapheme, graphemic, 12–15, 17, 18, 20, 22, 25–46, 106, 107, 115, 122, 123, 134, 138–140, 153 green, 64, 221, 229 halahama, 14–17, 20, 21, 25, 27–29, 40, 43, 46, 47 heavy lip sounds, see also labio-dentals, 87 hierarchy, 101 homonymy, 56 homophone, 183 homorganic, 94, 99 hyper-sanskritism, 157 iconic, 70, 105, 106 iconographical, 229 idiomatic, 58 illiterate, non-literate, see also literate, 2, 27, 54 imponderability, 12 ink, 31, 46 iron, 47 Iron Age, 33, 35, 37, 44, 51 island(s) Caroline Islands, 2, 3, 9, 10 Crete, 4 Cyprus, 18, 40, 47

Easter Island, 4 Maldive Islands, 9 Philippines, 6 Sai Island, 55–58, 62, 63, 66 Sri Lanka, 131 Tahiti, 70, 71, 78–82 isolation, isolated, 15, 114, 122, 169, 171, 190, 216 journal, 109, 112, 119, 121 journey, 73 king, see also emperor, ruler, president, 39, 208, 219 king Sargon, 184 king Sejong, 105, 127 kingdom, see also empire Kroraina Kingdom, 159 New Kingdom, 37 Northern Kingdom, 42 Que Kingdom, 45 labiodental, 87–89 labio-velar, 195 leaf, see palm leaf leather, 159 left-handed, 6 left-to-right, see also dextrograde, 18, 21, 101, 105, 106, 108, 110 ligature, 11, 12, 108, 119, 128, 136, 153, 156, 158, 160, 162 light lip sounds, see labio-dentals linear, 16, 18, 22, 25, 29, 31–34, 39, 40, 45, 105, 106 segmentallinear, 105 lingua franca, 41, 133, 150, 156, 163, 176 literate, literacy, see also illiterate, 1, 9, 27, 33, 39, 45, 54, 65, 129 loanword, loan, 2, 15, 21, 58, 59, 65, 98, 171, 177, 182, 183, 190, 191, 194, 196, 217 backloan, 196 logogram, 127, 181, 182, 184, 190, 191, 201, 207 logograph, logographic, 1, 4, 7–9, 105, 170, 171, 176, 178, 193, 196, 204 pseudo-logographic, 187 logophonographic, 169, 170 majuscule, see also minuscule, 108, 113, 116–119, 122, 123, 128 matres lectionis, 26, 33, 35, 46 minuscule, see also majuscule, 14, 15, 108, 113, 115–119, 122, 128 mirror, mirrored, 6, 89–92, 114 mission, missionary, 2, 66, 70, 78, 81, 124

subject index mistake, see also error, 15, 17, 20, 26, 69 monolingual, 201, 218 monophthong, monophthongization, 84, 107, 152 moon, 95 multilingual, see also bilingual, 46, 66, 95 multiple, 70, 175, 226 multiscriptual, see also biscriptual, 158 mute, 81 neo-syllabary, 23 newspaper, 6, 104, 105, 112, 116, 117, 126, 127 numeral, numerical, numerological, 2, 3, 5, 6, 17, 19, 23, 162, 181 obscure, obscured, 31, 73, 208 obsolete, 12, 91, 128, 137, 163 oral, 39, 44, 149, 171–174, 235 oralization, 34, 44 orange, 64 orthography, orthographic, 8, 9, 32, 34–36, 38, 42, 43, 55, 62, 65, 97, 100, 103, 106, 108, 109, 146, 148, 151, 156, 158–164, 181, 183–185, 187, 191, 201, 202, 206, 207, 210, 211, 214, 215, 219, 220, 222–237 ostracon, 16, 20, 32, 42, 137, 138, 141–144, 219, 228, 229 painted, 57 palaeography, palaeographical, 15, 21, 31, 32, 46, 151, 157 palimpsest, 21, 29, 50 palm leaf, 15, 157 paper, see newspaper papyrus, 31, 219–221, 225, 229, 233, 234, 236, 238, 239, 241–243 parchment, 6 phantom, 21 phonogram, phonographic, see also logophonographic, 98, 169–171, 176, 178, 181–184, 186, 187, 190–197, 201–211, 215, 216 play, 42 sistrum-play, 220 poem, poetic, poetry, see also rhyme, 20, 81, 169, 171, 179 politics, political, 46, 103, 123, 124, 126–128, 132, 163, 164 political science, 150, 151 socio-political, 65 polyphonic, polyphony, 30, 41, 47, 186, 215 polysemy, 56 polyvalence, 37, 38, 40 prescriptive, 95, 93

247

president, 123 prose, 169 prosodic, see also tone, 113 proverb, 57 pseudo, see logographic punctuation, 6, 94, 95, 102, 108, 113, 122 Que, see Kingdom of Que queen, 210 question mark, 2 rebus, 1, 182 reconstruction, reconstructed, 37, 131, 137, 138, 145, 151, 152, 183–185, 187, 190, 194–198, 201, 207, 216 reduction, 13, 18, 29, 30, 33 redundant, 202 regulated, see standardized rhyme, rime, 83, 85, 87, 90 right-to-left, see also sinistrograde, 17, 18, 21, 101, 108, 134 river Nile, 53, 54, 56, 57, 65 Tigris, 208 romantic, romanticized, 16, 31, 36, 57 romanization, romanized, Roman period, see also Roman, 53, 84, 100, 219, 224, 227, 236 ruler, see also emperor, king, president, 45, 134, 155, 190 Byblian ruler Zakarba, Zakurba, 38 Indian ruler Asoka, 133 Kerma rulers, 53 Ksatrapa ruler Damazada, 155 Luwian ruler Kulamu, 45 Rudradaman, 155 Scythian rulers, 154 sandhi, 26, 39, 151 scriptio franca, 13, 39, 41, 44, 45 sea, sea trade, 41, 45, 47 Dead Sea, 19 Mediterranean, 41, 44, 45 seal, 20, 49, 155, 156, 165 semiotic, 123 semisyllabary, 15 sideways, 124 signary, 23 silver, 18, 232 sinistrograde, see also right-to-left, 17, 18, 21 slang, 81 slanted writing, 121 slogan, 57

248

subject index

space syntax, 39, 49 spacing, 59, 62, 94–96, 104, 108, 111, 112, 122 spelling, see also orthography, 9, 26, 34, 43, 59, 70, 100, 103, 148, 151, 152, 156, 158, 183, 187, 188, 191, 195, 197, 198, 202, 209, 215, 230, 237 broken, 191–194, 195, 215, 216 magical, 137 resumptive, see broken standard, standardized, 16, 17, 19, 20, 27, 28, 34, 41, 43, 44, 45, 55, 75, 77, 79, 80, 87, 89, 99, 106, 108, 113, 120, 148, 160, 162, 164 star, 24, 182 stela, 219, 220, 228 stenography, 113 stone, 6, 14, 15, 19, 31 stress, 55, 234 stroke, 89, 94–96, 99, 107, 114, 153, 161 dividing, 21 double, 99 horizontal, 107, 109, 111–114, 162 slanting, 21 substroke, 139 vertical, 107, 111, 112 sun, 95 sun-disk, 231 sun god, 207 syllabogram, 95, 98, 128 tablet, see also clay, wooden, numerical, 18, 137, 138, 141–144, 161, 181, 190, 207, 219, 227, 228, 230

tautosyllabic, 93 theonym, see God tone marking, see also prosodic, 56, 94, 98, 99 top-down, top-to-bottom, 93, 108, 124 toponym, geographical name, 38, 176 trilingual, see also multilingual, 101 truck, pick-up truck, see also vehicle, 57, 58, 63 typewriting, typewriter, 124, 125, 128, 129 unetymological, 9, 220–226, 228, 230–237 vehicle, see also buksi, car and truck, 55, 57, 59, 65, 152 versatility, 9 white, 208 white space, 95, 96 wine, 42, 219 wood, wooden, 15, 137, 159, 219 word play, see play, games world, 11, 30, 56, 72, 104, 105, 124, 127 Islamic world, 11 world order, 104 yellow, 198 Zakarba, Zakurba, see ruler zero, see also numeral, 90, 108, 121, 124, 192, 193, 236

LANGUAGE (GROUP) AND SCRIPT INDEX Afro-Asiatic, 185 Akhmimic, see also Akhmim, 232 Akkadian, see also Sargonic Akkadian, Ebla Akkadian, 9, 181–218 Akkade, 192, 198 Ammonite, 20 Arabic, 3, 5, 8, 11–15, 23, 26, 28, 30, 33, 46, 48, 49, 53–67, 185, 186 Aramaic, 20, 26, 29, 30, 35, 38, 40, 42, 43, 45, 46, 133, 186 Austronesian, 78

Finnish, 77 French, 75–77, 79, 81, 82 Futu, 79

Babylonian, 184, 185, 190, 196, 197, 208, 216 Bactrian, 155, 164 Bamum, 2 Bohairic, 221, 232 Brahmi, 8, 131–168 Braille, 5 Brazilian Portuguese, 75

Han’gŭl, 83–85, 90, 94–97, 99, 100, 103–129 Hebrew, proto-Hebrew, 16, 19, 20, 26, 29, 30, 35–43, 46 Hieratic, 9, 53, 219, 220, 223–225, 227, 233, 234 Hieroglyphic, 4, 53, 54, 219–221, 224, 227, 228, 230, 231, 235 Hiragana, see also Kana, 169, 170 Hittite, 185 Hittito-Mesopotamian, 22 Hurrian, 185

Canaanite, 18, 20, 31, 32, 34, 36–40, 43–45 Carian, 5, 9 Caroline Islands script, 2, 3, 9 Cherokee, 1, 3, 5 Chinese, 1, 3, 4, 8, 54, 83–105, 121, 123, 124, 127, 128, 137, 163, 169–178 Coptic, Old Coptic, 53, 54, 221, 223, 227, 231, 238 Cuneiform, 1, 2, 4, 9, 16-18, 20–22, 27, 30–32, 40, 54, 181–217 Cyrillic, 2, 3, 73, 76, 106, 116, 128 Demotic, 4, 9, 53, 219–243 Devanagari, 104 Drehu, 78–82 Ebla Akkadian, EbAkk, 183, 185, 201–210, 215, 216 Ebla Sumerian, EbSum, see also Sumerian, 206, 208–210 Egyptian, 1, 2, 4, 9, 15, 24, 32, 37–39, 49, 53, 54, 104, 219–243 English, 2, 3, 54, 69, 71, 79–81, 88, 89, 96–98, 108, 124, 164, 173 Ethiopic, 15, 23, 186 Etruscan, 73 Fayumic, 227, 232

Gāndhārī, 132–134, 136, 145–148, 150, 151, 155–159, 162, 163 Ge’ez, 15 German, 12, 69 Greek, Old Greek, 5, 8, 12, 20, 23–26, 30, 35, 40, 46, 53, 54, 66, 69, 70, 72–77, 80, 104, 133, 154, 155, 225, 234, 237–240

Indian, see Brahmi, Devanagari, Kharoṣṭhī Indo-Aryan, 167 Indo-European, 25, 26 Indo-Greek, 154 Indo-Scythian, 154 International Phonetic Alphabet (IPA), 13, 121 Iranian, 160 Israelite, 20, 35, 42 Italian, 75 Japanese, see also Kanji, Kana, 3, 8, 57, 83–86, 88, 93, 103, 120, 169–179 Sino-Japanese, 97, 103, 169–171, 176–178 Judahite, 35, 42 Kana, 95, 100, 103, 169–171 Kanji, 169, 170, 178 Katakana, see also Kana, 169 Kharoṣṭhī, 8, 131–168 Korean, see also Han’gŭl, 8, 83–129 Sino-Korean, 85, 86, 93, 97, 105, 109, 121, 123, 126 Kufijic, 62

250

language (group) and script index

Latin, 15, 16, 53, 79–82, 104, 172, 176, 219 Linear B, 4

Russian, 2, 69, 70, 75–77, 81, 82 Ryūkyūan, 83, 84, 88

Mahas, see Nubian Maldivian, 3, 5 Manchu, 84, 85, 88, 93–101, 128 Meroitic, 4, 53, 54 Mindoro, see Philippine scripts Moabite, 35 Mongolian, 84, 91, 93–97, 100, 101, 128 Morse code, 5

Sabaic, Sabean, Sabaean, 14, 25, 28, 43 Saidic, 221, 232 Sanskrit, 8, 90, 91, 104, 131–164 Sargonic Akkadian, SargAkk, 183–185, 187–191, 193, 201–204, 214–216 pre-Sargonic, PreSargSum, 183, 184, 187, 189, 190–196, 215, 216 secret script, 5 Semitic, 185, 186, 194, 236, 240 Proto-Semitic, 185 Sinaitic (Proto-), 21, 31, 32 Sino-Japanese, see Japanese Sino-Korean, see Korean slang, 81 South Arabic (Old), see Sabean Spanish, 69, 74–76 Sudanic, 53 Sumerian, see also Ebla, 9, 181–184, 187, 190–197, 201, 206–210, 214–216 Syriac, 29

Naskhi, 62 Nengone, 78, 80–82 Nilo-Saharan, 53 Nobiin, see Nubian Nubian (Old), 8, 53–67 North Eastern Sudanic, see Sudanic Oceanian, see also Polynesian, 78 Ogham, 5 Palmyrene, 29 Persian (Old), 1, 3 Phagspa, 90, 91 Philippine script, 6 Polynesian, 71, 78 Portuguese, 75 Ptolemaic, see also Egyptian, 53, 219, 223, 227 Punic, see Phoenician Roman, 3, 8, 54, 57, 73, 103–108, 111, 114–116, 121–123, 125, 127, 128 Rongorongo, 4

Tahitian, see also Polynesian, 70, 71, 78–81 Thaana, see Maldivian Tokharian, 159 Ugarit, 15–31, 35, 39–43, 185, 186 Uruk, 181, 182, 201, 206 Vedic, 131 Vietnamese, 172 Wallis, 79

AUTHOR INDEX Abel-Rémusat, J.-P., 90, 101 Adiego, I.J., 5, 9 Ahonen, L., 77, 82 Albright, W.F., 21, 29, 47 Allchin, R.R., 165 Allon, M., 132, 157, 159, 160, 164 Amadasi Guzzo, M.G.A., 33, 39, 48 Aman ur Rahman, see Rahman An, Cha-san, 104, 129 Arakin, V.D., 71, 82 Archi, A., 211, 217 Artzy, M., 41, 47 Assmann, J., 24, 26, 47 Attinger, P., 194, 195, 217 Baines, J., 50, 167 Banaschak, S., 229, 240 Barsky, R.F., 34, 47 Batt, C.M., 165 Bauer, J., 182, 217 Baums, S., 132, 134, 147, 148, 164 Baurain, Cl., 48, 51, 52 Bautze-Picron, C., 165 Bechtolsheim, P. von, 217 Beeston, A.F.L., 15, 49 Bennet, J., 50, 167 Bentley, J.R., 179 Bernal, M., 12, 47 Beyer, K., 35, 47 Bhandare, Sh., 166 Bhattacharya, G., 155, 156, 165 Black, J., 217, 218 Blanco, A.G., 52 Blau, J., 30, 35, 40–42, 47 Bonaparte, R., 91, 101 Bonnet, C., 48, 51, 52 Bopearachchi, O., 155, 164 Bordreuil, P., 17, 21, 47 Borger, R., 183, 217 Boschung, D., 47 Boyer, A.M., 159, 161, 165 Braarvig, J., 164 Brashear, W., 240 Bright, W., 2, 9, 23, 48, 50, 136, 165 Brody, A.J., 41, 47 Bron, F., 15, 47 Brønno, E., 35, 47 Brough, J., 137, 145, 159, 163, 165

Burrow, Th., 159, 161, 165 Byrne, R., 20, 47 Callieri, P., 167 Capasso, M., 243 Casaretto, A., 24, 47 Ch’oe, Hyŏn-bae, 115, 116, 117, 119, 123, 129 Chattopadhyaya, B.D., 155, 165 Childe, V.G., 44, 45, 48 Chosŏn Chŏngmunhoe, 109, 113, 115, 116, 129 Chosŏnŏ Hakhoe, 106, 119, 120, 122, 129 Chu, Si-gyŏng, 104, 109, 112, 122, 129, 130 Chung Kwang, 99, 101 Chung Kei Won, 104, 129 Civil, M., 183, 190, 217 Clarysse, W., 243 Clauson, G., 91, 101 Cochavi-Rainey, Z., 35, 37, 39, 51 Cohen, Ch., 49 Congrès international des Orientalistes, 100, 102 Coningham, R.A.E., 131, 165 Conklin, H.C., 6, 9 Coulmas, F., 15, 26, 30, 48 Creason, S., 30, 48 Cross, F.M., 19, 21, 48 Crum, W., 238, 240 Crystal, D., 25, 48 Cunningham, A., 155, 165 Cushman, E., 3, 9 Damsteegt, Th., 150, 165 Dani, A.H., 136, 165 Daniels, P.T., 2, 7, 9, 22–26, 48, 50, 105, 129, 136, 165 Daues, A., 24, 47 Davoli, P., 243 Degen, R., 35, 48 Depauw, M., 242 Depuydt, L., 238, 240 DeSilva, M.W.S., 3, 10 des Places, see Places Deutsch, R., 38, 48 Deutscher, G., 218 Devauchelle, D., 242 Diakonofff, I.M., 195, 217 Diehl, J.F., 49

252

author index

Dietrich, M., 16–18, 29, 31, 48 Drewes, D., 150, 165 Dücker, B., 242 Edzard, D.O., 207, 217 Englund, R.K., 181, 182, 217 Ernst, W., 241 Faber, A., 185, 186, 217 Fairbank, J.K., 179 Falk, H., 131, 133, 139, 155–158, 160, 161, 165–167 Farber, W., 217 Finkel, I.L., 7, 9, 10, 62, 67, 242 Fitzenreiter, M., 241 Folmer, M.L., 40, 43, 48 Freedman, D.N., 19, 48 Friedrich, J., 33, 39, 48 Fussman, G., 137, 157, 158, 166 Ganor, S., 32, 50 García-Ramón, J.L., 24, 47 Garfijinkel, Y., 32, 50 Garg, S., 166 Garr, W.R., 35, 39, 43, 48 Gelb, I.J., 22, 23, 48, 179, 186, 187, 217 George, A.R., 199, 217 Gerhard, V., 24, 48 Getatchew Haile, 15, 48 Gitin, S., 47, 51 Glass, A., 131, 132, 138–140, 145, 147, 148, 153, 160–162, 164, 166 Goodspeed, E.J., 225, 241 Gordon, A.E., 73, 82 Goyon, J.C., 227, 241 Gragg, G.B., 194, 217 Greenfijield, J., 41, 48 Grenfell, B.A., 225, 241 Grevisse, M., 75, 82 Grifffijith, F.Ll., 53, 67, 228, 241 Gros, M., 49 Grothofff, Th., 229, 240 Gruendler, B., 13, 15, 48 Günther, H., 50, 52 Guzzo, see Amadasi Gzella, H., 19, 49, 50 Haase, W., 240 Habein Yaeko Sato, 179 Haider, P.W., 41, 49 Haile, see Getatchew Halayka, I.K.H., 44, 49 Hallo, W.W., 19, 49 Hartmann, J.-U., 166, 167

Hāshim, see Muhammad Hasselbach, R., 184–189, 217 Hattori Shirō, 88, 101 Havelock, E.A., 237, 241 Hawley, R., 16, 17, 49 Heide, M., 20, 49 Hellenkemper, H., 47 Heltzer, M., 38, 48 Higuchi Kōichi, 90, 91, 101 Hilgert, M., 197, 216, 217 Hinüber, O. von, 131, 149, 166 Hoch, J., 35–39, 49 Hofffmann, F., 220, 221, 226, 227, 230, 233, 241, 243 Hofmann, J.B., 15, 16, 52 Hohlfelder, R.L., 47 Hope, E.R., 90, 91, 102 Houston, S., 50, 167 Hunt, S., 225, 241 Hurvitz, A., 49 Irvine, A.K., 15, 49 Iwamatsu, A., 159, 166 Jacobsen, G., 157, 159, 164 Jagersma, A., 195, 217 Janowski, B., 242 Jasnow, R., 220, 241 Jefffery, L.H., 12, 49 Junast, 91, 101 Kaneshiro, S., 2, 10 Kangle, R.P., 152, 166 Kerr, R.M., 13, 49 Khan, G., 49 Kiefffer-Pülz, P., 167 Kim-Renaud, Y.-K., 129 Kim, Chin W., 106, 129 Kim, Chŏng-su, 109, 113, 118, 120, 121, 123, 125, 129 Kim, Min-ju, 129, 130 Kim, Min-su, 109, 110, 123, 129 Kim, Sŏk-kon, 117–119, 123, 124, 129 Kim, Tu-bong, 106, 108, 113, 114, 122, 129 Kim, Yun-gyŏng, 116, 118, 120, 129 King, R., 103, 104, 109, 110, 112, 129 Kittler, F., 241 Kloekhorst, A., 185, 217 Knauf, E.A., 40, 49 Knuf, H., 241, 242 Kōichi, see Higuchi Koizumi, Y., 138, 166 Kong, Pyŏng-u, 124, 125, 128, 129 Konow, S., 155, 156, 160, 166

author index Kooij, G. van der, 22, 31, 46, 49 Kouwenberg, N.J.C., 185, 217, 218 Krahmalkov, C.R., 33, 37, 49 Krebernik, M., 33, 41, 49, 182, 217 Krings, V., 48, 51, 52 Krispijn, Th.J.H., 190, 191, 195, 214, 215 Kwang, see Chung Kwang Laroche, E., 185, 218 Larson, J.A., 242 Ledyard, G.K., 90, 91, 102 Lee Ki-Moon, 97, 102 Lehmann, R.G., 20, 31, 32, 39, 40, 49, 50 Leitz, Chr., 229, 241, 242 Lemaire, A., 27, 30, 32, 33, 41, 45, 50 Lenormand, M.-H., 81, 82 Lenz, T., 161, 166 Lippert, S.L., 220, 241, 243 Loretz, O., 16–18, 29, 31, 48 Loundine, A.G., see Lundin Lubetski, M., 49 Lucy, D., 165 Lüders, H., 156, 166 Ludwig, O., 50, 52 Lundberg, M.J., 19, 52 Lundin, A.G., 16, 17, 31, 50 Luther, A., 49 Macdonald, M.C.A., 15, 27, 33, 46, 50 Malinar, A., 150, 166 Martin, M.F., 21, 29, 50 Masayoshi, see Shibatani Mayer, W.R., 48 Mazar, A., 51 McCarter, P.K., 51, 52 Meister, G., 88, 102 Michalowski, P., 183, 218 Millard, A., 18, 34, 44, 50 Miller, C.L., 9, 48, 50 Miller, R.A., 179 Milne, A.A., 78, 82 Misgav, H., 32, 50 Monroe, C.M., 41, 45, 50 Muḥammad Jalāl Aḥmad Hāshim, 54, 67 Müller, A., 15, 50 Müller, W.W., 88, 102 Nàisī, see Yáng Nam Kwangu, 86, 102 Naveh, J., 46, 50 Nebes, N., 15, 50 Negbi, O., 41, 50 Nissen, H.J., 181, 217 Noble, J., 159, 161, 165

253

Noble, P.S., 165 O’Connor, M., 25, 50 Ohlig, K.-H., 49 Olson, D.R., 179 Omar Al-Ghul, 49 Osing, J., 228, 234, 238, 241, 242 Overmeer Fisscher, J.F. van, 88, 102 Pagel, U., 137, 166 Pai, Hyung Il, 129 Pak, Hwan, 112, 130 Pardee, D., 16–19, 28, 29, 47, 50 Paul, Sh.M., 49 Pettinato, G., 206, 207, 211, 218 Philippi, D.L., 179 Places, E. des, 240, 241 Plato, 69, 82 Powell, B., 22, 24, 50, 237, 241 Pratt, K., 124, 130 Pries, A., 221, 241 Puech, E., 16, 50 Quack, J.F., 15, 38, 51, 219–221, 223–225, 227, 228, 235, 236, 238, 240–242 Raddatz, F.M., 24, 26, 47 Rahman, A., 156, 165 Rainey, A., 33, 36, 51 Rapson, E.J., 139, 165 Recklinghausen, D. von, 241, 242 Redford, D.B., 52 Reischauer, E.O., 179 Renz, Th., 16, 51 Ri, P’il-su, 114, 122, 130 Riesenberg, S.H., 2, 10 Rilly, C., 53, 67 Robin, Ch., 15, 17, 47, 51 Roche, C., 49, 51 Roeder, H., 242 Röllig, W., 17, 33, 39, 48, 51 Rollinger, R., 49 Rollston, Ch., 19, 31, 32, 46, 51 Roth, M.T., 217 Rubio, G., 201, 205, 218 Rutt, R., 124, 130 Ryckmans, J., 15, 51 Ryholt, K., 230, 241, 242 Sadakata, A., 158, 167 Salomon, R., 132, 137, 138, 150, 151, 154, 156–161, 164 Sander, L., 161, 167 Sanders, S.L., 19, 20, 51, 218

254

author index

Sansalvador, A.V., 24, 47 Sarbaugh, J., 3, 10 Sass, B., 31, 32, 34, 51 Sato, see Habein Saur, M., 41, 51 Schami, R., 11, 12, 51 Schentuleit, M., 220, 241, 243 Schipper, B.U., 38, 39, 41, 51 Schmitz, Ph.C., 39, 51 Schneider, Th., 37, 38, 51, 241 Schoors, A., 243 Schulz, H., 32, 51 Sederholm, V.H., 238, 239, 240, 242 Seeley, Chr., 179 Segert, S., 185, 218 Seiji, see Tsuji Sejong Taewang Kinyŏm Saŏphoe, 124, 125, 130 Senart, E., 159, 161, 165 Senior, R., 155, 167 Sherratt, A., 41, 51 Sherratt, S., 41, 51 Shibatani Masayoshi, 179 Shin’ichirō, see Tawata Shisha-Halevy, A., 238, 242 Shirō, see Hattori Siebold, Ph.F. von, 90, 102 Simson, G. von, 156, 167 Sivan, D., 21, 35, 37, 39, 51 Smith, D., 18, 51 Smith, M., 219–221, 223, 224, 226, 227, 229, 242 Sollberger, E., 196, 218 Sommerfeld W., 182, 218 Stadler, M., 222, 224, 242, 243 Starcky, J., 38, 52 Stein, P., 15, 52 Steiner, R.C., 35, 40, 52 Stern, E., 47, 51 Stolper, M.W., 217 Strauch, I., 131–133, 138, 150, 151, 157, 158, 166, 167 Sürenhagen, D., 181, 218 Swiggers, P., 23, 33, 41, 52 Swiny, H.W., 47 Swiny, S., 47 Tandon, P., 155, 167 Tangherlini, T.R., 129 Tappy, R., 19, 51, 52 Tawata Shin’ichirō, 88, 102 Teeters, E., 242 Thissen, H.J., 243 Thompson, H., 228, 241

Tropper, J., 17, 18, 21, 27, 50, 52 Tsakos, A., 67 Tsakos, H., 67 Tsuji Seiji, 96, 102 Ullendorfff, E., 40, 52 Vennemann, Th., 14, 52 Vergote, J., 238, 243 Vertogradova, V.V., 157, 158, 161, 168 Villani, P., 179 Vita, J.-P., 18, 52 Vleeming, S., 219, 220, 243 Vogel, J.Ph., 155, 168 Voigt, R., 133, 168 Volten, A., 238, 243 Voogt, A.J. de, 2, 3, 5, 7, 9, 10, 62, 65, 67, 242 Vovin, A., 179 Wachter, R., 40, 52 Wäfler, M., 217 Walde, A., 15, 16, 52 Walker, W., 3, 10 Wall, C.W., 90, 102 Wehmeyer, A., 179 Westenholz, A., 187, 215, 218 Wevers, J.W., 35, 40, 52 Whittaker, M., 24, 52 Widmer, G., 219–221, 227, 228, 232, 242, 243 Wiesehöfer, J., 49 Wilhelm, G., 242 Willems, H., 243 Witte, M., 49 Woods, C., 183, 218 Woodward, R.D., 48, 50 Wylie, A., 91, 102 Yáng Nàisī, 91, 101 Yi Kimun, see Lee Ki-Moon Yi, Kyu-yŏng, 108, 112, 122, 130 Yŏnhŭi taehakkyo tongbanghak yŏn’guso (YTTY), 97, 102 Young, I., 45, 46, 52 Yule, Henry, 91, 102 Zamora, J.A., 52 Zauzich, K.-Th., 220, 236, 241, 243 Zhōngguò mínzú gŭwénzì yánjiūhuì, 101 Ziyadeh, A., 49 Zólyomi, G., 217, 218 Zoppi, U., 164 Zuckerman, B., 52