=Paper=
{{Paper
|id=Vol-3232/paper22
|storemode=property
|title=Encoding Hieroglyphic Texts
|pdfUrl=https://ceur-ws.org/Vol-3232/paper22.pdf
|volume=Vol-3232
|authors=Heidi Jauhiainen
|dblpUrl=https://dblp.org/rec/conf/dhn/Jauhiainen22
}}
==Encoding Hieroglyphic Texts==
https://ceur-ws.org/Vol-3232/paper22.pdf
Encoding Hieroglyphic Texts
Heidi Jauhiainen1
1
University of Helsinki, P.O. Box 4, 00014 University of Helsinki, Finland
Abstract
With the help of data science, researchers in the humanities can study large amounts of data at once and
find regularities that they might not otherwise detect. In order to use digital methods, the texts to be
examined must be in machine-readable form, but the lack of such text corpora hinders the digital study
of ancient Egyptian texts. A sign can be next to, above, or over another in a hieroglyphic text, and two or
more signs can be nested. Egyptologists use encoding to maintain the information on the signs and their
places relative to each other when preparing hieroglyphic texts for publication in printed form. The
encoding uses letter-number combinations from the Gardiner list, a standard reference list for Ancient
Egyptian hieroglyphs. To increase the number of machine-readable hieroglyphic texts, the plan is to
develop a workflow that uses automatic transliteration. This paper aims to present the first steps towards
this goal. Ancient Egyptian texts are encoded by hand in JSesh, an open-source hieroglyphic editor. The
aim is to publish annotated texts in a structured form, and a tool is being built to turn the binary format
files produced in JSesh into machine-readable form. This paper introduces Gly2Mdc version 1.0, which
extracts and cleans the encoding from the binary file. The tool is openly available and can be used for
files with the extension .gly and containing encoded hieroglyphic text.
Keywords
Encoding, hieroglyphic texts, tool
1. Introduction
To use digital methods for researching texts, the texts must be machine-readable. For major
modern languages, such as English, there are openly available digital corpora of texts which
can be extensive and built from natively digital texts, such as Wikipedia. A smaller corpus
is sufficient for many purposes, and there are several corpora built specifically for historical
research. Assyriology, for example, has freely downloadable corpora of machine-readable
cuneiform texts, such as Open Richly Annotated Cuneiform Corpus.1 However, the lack of
similar corpora hinders the digital study of ancient Egyptian texts.
The aim of the project Machine-Readable Texts for Egyptologists 2 is to develop a workflow
for producing machine-readable hieroglyphic texts. OCRing hieroglyphic texts would produce
machine-readable texts, but training the method would require a lot of annotated texts in the
same handwriting; annotated texts that are currently not available. Using Unicode characters
to build machine-readable hieroglyphic texts would also be an option, as there are over 1,000
The 6th Digital Humanities in the Nordic and Baltic Countries Conference (DHNB 2022), Uppsala, Sweden, March 15-18,
2022
Envelope-Open heidi.jauhiainen@helsinki.fi (H. Jauhiainen)
Orcid 0000-0002-8227-5627 (H. Jauhiainen)
© 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings CEUR Workshop Proceedings (CEUR-WS.org)
http://ceur-ws.org
ISSN 1613-0073
1
http://oracc.museum.upenn.edu
2
https://blogs.helsinki.fi/hwikgren/marete/
244
hieroglyphic Unicode characters.3 Unfortunately, the hieroglyphic block is outside the Basic
Multilingual Plane, and the characters are not correctly handled by commonly used software
applications such as Microsoft Word.
In a hieroglyphic text, the signs can be next to, above, or over another, or even nested. There is,
therefore, a tradition of using encoding to maintain the information on the signs themselves and
their places relative to each other when preparing hieroglyphic texts for printed publications.
However, the encoded texts are usually discarded after the text has been built and turned into
a picture. Egyptologists are, in fact, not used to handling hieroglyphic texts in encoded form.
Instead, they transliterate hieroglyphic texts with Latin letters and diacritics as preparation for
translating and analyzing them. Computer-assisted transliteration of hieroglyphic texts will
speed up producing such texts in machine-readable form.
This paper presents the first stages of building an automated transliteration method for
hieroglyphic texts and the first version of a tool to help in the process. To build the method, one
needs machine-readable hieroglyphic texts. Since there is no working automated method for
producing these, the chosen method is to encode texts by hand in a hieroglyphic editor called
JSesh [1]. The encoded texts will be annotated with the automatically produced transliterations
and eventually published in a structured form. For this end, a tool called Gly2Mdc4 is being
built. Version 1.0 of the Java-based tool handles the cleaning and writing the encoding from
binary form to text file, a process that would otherwise be tediously slow.
I will first briefly describe the intricacies of hieroglyphic writing (Section 2) and why encoding
is at the moment the best way of producing machine-readable hieroglyphic texts instead of, for
example, Unicode fonts (Section 3). I will then outline the use of JSesh for producing encoded
hieroglyphic texts (Section 4) before introducing version 1.0 of the Gly2Mdc tool for producing
structured annotated files for publishing the machine-readable texts generated in the project
(Section 5).
2. Hieroglyphic Text
Hieroglyphs were generally used in monumental texts. They could be written in different
directions, from right-to-left, left-to-right, or in vertical columns continuing either right or
left. The direction of a text usually depends on the related pictures. One can tell the writing
direction from the living creatures amongst the hieroglyphs; one always approaches people and
animals from the front.
When writing literature, letters, or administrative texts, ancient Egyptians generally used
so-called hieratic writing, which is cursive handwriting that is based on hieroglyphic writing
(see Figure 1). In the non-literary documents of the New Kingdom period (c. 1550-1069 BCE),
hieratic was usually written from right to left.
Different hieroglyphic graphemes, generally called signs, have different functions, and one
sign can have varying functions in different contexts [2]. Signs can be used as logograms,
referring to entire words. Many signs function as phonograms representing one to three, rarely
four, phonemes. Ancient Egyptian is an Afro-Asiatic language and, just as in Semitic languages,
3
https://unicode.org/charts/PDF/U13000.pdf
4
https://github.com/MaReTEgyptologists/gly2mdc
245
Figure 1: A hieroglyphic text written in hieratic handwriting on a piece of pottery. O. Turin N 57458,
from Deir el-Medina, New Kingdom (c. 1550-1069 BCE). In Museo Egizio, Turin.
only consonants and semivowels were represented in writing. Both logograms and phonograms
could be used to write words that had nothing to do with the sign they depict. Some signs
have several different phonetic values, and several signs could have the same phonetic value.
Therefore, signs were combined, and interpretants, or phonetic complements, were used to
show which phonetic value was meant (see Figure 2). In order to tell the different words with
the same phonetic values apart, classifiers were added to the end of the word.
Figure 2: The sign depicting a chisel that has two phonetic values ab (with Egyptological aleph) and mr
is here used with phonetic complements and classifiers in different words. C stands for a classifier.
A Hieroglyphic text does not indicate word boundaries nor the end of a sentence. Furthermore,
a hieroglyph was often written above or over another one, or they could be nested. Hieroglyphic
words could also be written in multiple ways depending on the space and aesthetic preferences
of the scribe. Producing machine-readable hieroglyphic texts is, hence, not straightforward.
3. Machine-Readable Hieroglyphic Texts
One way of producing machine-readable hieroglyphic texts would be to use Unicode, which,
since 2009, includes over 1000 hieroglyphic signs. In 2019, the so-called format control characters
for positioning the hieroglyphic signs were introduced in Unicode version 12, and, since then,
246
it has been possible to position the signs properly. However, writing hieroglyphic texts with
Unicode code points—for example, 13001 for the sitting man in Figure 2—is not easy. There
are fonts for Unicode hieroglyphs, but they require specific tools depending on whether one
wants to use them, for example, on a web page or a document. One cannot just download the
font and start writing hieroglyphs in a text file. Moreover, none of the available hieroglyphic
keyboards seem to provide the possibility of positioning the signs using the newly introduced
format control characters.5
Transliteration refers to the conversion of text written using one writing system to another,
and Egyptologists use it to transform texts written in hieroglyphs into words written in Latin
letters and diacritical marks. The various transliteration fonts for ancient Egyptian texts are far
easier to use than Unicode characters for Egyptian hieroglyphs. Hence, producing transliterated
hieroglyphic texts in machine-readable form would be possible. However, the conversion
is more complex than in many other languages, as the writing system uses characters that
correspond to more than one Latin letter or have alternatives for equivalences. Because many
hieroglyphs, particularly the groups formed by them, can be transliterated in many ways, manual
transliteration of a hieroglyphic text requires examination of vocabularies and conclusions
drawn from that, making it a slow process. Transliteration is already an interpretation of the
text, and, more importantly, it does not retain the information on the hieroglyphic signs used.
In order to display the signs properly in printed books and online texts, the hieroglyphic
texts are encoded with Latin letters and Arabic numbers (see Figure 3). The codes come from
the so-called Gardiner Sign List,6 the standard reference list for Egyptian hieroglyphs. The list
was compiled by Sir Alan Gardiner in 1927 for his grammar of the Middle Egyptian language
[3], and the letters refer to various categories of signs while each sign has a number within its
category.
Figure 3: The first words of the text in Figure 1 transcribed from the cursive original to hieroglyphs
with Manuel de Codage (MdC) encoding. The line with the transliteration of the signs displays the
alternatives when available. C stands for a classifier. A person with the name Amen-Khau is swearing
an oath.
There are a few different methods of encoding hieroglyphs [4], the most used of which is
called Manuel de Codage (MdC) [5]. To place a sign above another, MdC uses a colon, and
an asterisk is used to indicate that signs are next to each other [6]. However, MdC does not
indicate different sizes of signs, and there is no way of placing a sign over another or nesting
them. MdC is used in various hieroglyphic editors that are often operating system specific, and
5
E.g. Keyman Hieroglyphic keyboard https://help.keyman.com/keyboard/hieroglyphic/1.4/hieroglyphic.
6
https://en.wikipedia.org/wiki/Gardiner's_sign_list
247
each software has dealt with these shortcomings differently [7].
Mark-Jan Nederhof proposed a new encoding system which he called Revised Encoding
Scheme (RES) [5]. With RES, signs can be nested and positioned over each other in multiple ways.
The position is not absolute; instead, it is defined in relation to other signs in the group. RES has
not been widely adopted by Egyptologists [8]. Because of its precise and lengthy commands,
they find RES too slow to write. It is also considered unnecessary since Egyptologists are
accustomed to the graphical hieroglyphic editors using MdC.
The encoded hieroglyphic texts are machine-readable, and Egyptology has, thus, a tradition
of producing machine-readable texts. Unfortunately, the encoding has only been considered
an intermediate step when publishing the texts as pictures in books and articles. There is no
tradition of publishing the encoding of the texts, and, instead, these are often discarded [7].
4. Encoding with JSesh
In the Machine-Readable Texts for Egyptologists project, I have chosen to encode hieroglyphic
texts with MdC using the software called JSesh [1]. JSesh is an open-source, Java-based,
operating system independent word processor for producing hieroglyphic texts. In order to deal
with the shortcomings of MdC, JSesh uses some additions to the encoding scheme. For example,
one can use the ’##’-code to place a sign over another. ’&’ and ’^^^’ can be used to group
and nest signs, respectively. There are also prefabricated composite signs, and it is possible to
position and resize signs manually. The texts are saved as binary files (.gly) compatible with
other hieroglyphic editors, such as Winglyph and Tksesh. The texts produced can be exported
as pictures, PDFs, or in RTF format, and it is possible to copy out the MdC encoding or the text
as Unicode characters.
Since the aim is to produce machine-readable texts, not pictures, the attention is on the
MdC encoding. The texts will be published for future use in a structured and annotated format
using TEI markup language,7 the structure preferred in Egyptology [9]. In order to be able
to build the TEI formatted files correctly, it is essential that the MdC encoding is as clean as
possible. Resizing and positioning signs manually in JSesh clutters the encoding. For example,
it is possible to nest sign D46 inside sign I9 by hand, but the encoding will end up looking
like this: I10\81**D46{{36,525,63}} while I10&D46 would require less complicated rules when
analyzing the encoded texts digitally. Thence, no manual positioning and resizing will be done
to the texts.
In the future, operating hieroglyphic Unicode characters might be more straightforward and,
hence, it is considered beneficial to annotate the texts with Unicode characters in addition to
the transliteration. Therefore, care is taken not to use signs or codes that are not in the Unicode
chart for hieroglyphs,8 although additional sign-code combinations are available in JSesh.
7
Text Encoding Initiative, https://tei-c.org
8
https://www.unicode.org/charts/PDF/U13000.pdf
248
5. Gly2Mdc
In JSesh, there is no option of saving the encoding directly to a text file; instead, it has to be
copied by hand to a new file. This operation requires many steps even when the chosen default
copy mode is the encoding: 1. select all the text, 2. copy, 3. open a new file in a text editor, 4.
paste, and 5. save the new file under a chosen name. To produce the TEI formatted files that
contain the annotated MdC encoding, it is thus, best to export the encoding directly from the
binary files with the extension .gly. For this purpose, I have built a tool called Gly2Mdc.
Gly2Mdc version 1.0 extracts the MdC encoding from the binary file, cleans the text, and
writes it to a new text file. The binary file contains information on the settings used when
producing the hieroglyphic text in the editor. These extra lines are removed during the cleaning
process. In the MdC encoding, a code is separated from other codes with a hyphen unless
the signs are marked as forming a sign combination. In JSesh, the codes are separated with
an underscore-hyphen combination (e.g., O1_-D21:X1*N5_-A24). The underline and hyphen
are replaced with a single white space during the cleaning to make the reading and future
processing of the encoding easier.
Gly2Mdc has been built using Java, and both an executable jar file and the source code are
openly available on GitHub with some sample texts.9 The tool takes the binary file produced with
JSesh as input and produces the encoded text to a file with the same name but extension .gly2mdc.
It works with a simple command java -jar Gly2mdc.jar , where
’binary_file_name’ is the name of the file containing the text produced with JSesh. In the future
versions of Gly2Mdc, the tool will be expanded to writing the annotated text using TEI markup
language.
6. Conclusions
Gly2Mdc can be used by anyone producing hieroglyphic texts with JSesh and other hieroglyphic
editors that save the files in binary format with the extension .gly. Instead of being thrown
away after the texts has been saved as a pictures or PDFs, the binary files can be turned into
text files with the encoded texts in machine-readable form for others to use.
Since JSesh version 7.5.5 was released in 2020, it is possible to copy the text in Unicode
characters, but that takes as many steps as copying the MdC. More, in fact, since only MdC can
be set as the preferred clipboard format and copying Unicode requires going through the Edit
menu and choosing which format to copy. In order to bypass this tedious process, Gly2Mdc
will be expanded to add the Unicode character to the metadata of each sign. A possibility of
writing the text to a text file in Unicode hieroglyphs is also planned.
Acknowledgments
I gratefully acknowledge the funding received from the Finnish Cultural Foundation and the
Kone Foundation that has made this research possible. Furthermore, I would like to extend my
gratitude to Dr. Tommi Jauhiainen for reading and commenting on the draft of the article. I
9
https://github.com/MaReTEgyptologists/gly2mdc
249
am also grateful to assistant professor Saana Svärd for including me in the inspiring research
collective of the Academy of Finland Center of Excellence in Ancient Near Eastern Empires.
Finally, I would like to thank the anonymous reviewers for their valuable comments.
References
[1] S. Rosmorduc, JSesh Documentation. [online], 2014. URL: http://jseshdoc.qenherkhopeshef.
org.
[2] S. Polis, The Functions and Toposyntax of Ancient Egyptian Hieroglyphs: Exploring the
Iconicity and Spatiality of Pictorian Graphemes, Signata: Annales des Sémiotiques 9 (2018).
doi:10.4000/signata.1920 .
[3] Sir A. Gardiner, Egyptian Grammmar: Being an Introduction to the Study of Hieroglyphs,
3rd. ed., Griffith Institute, Oxford, 1957.
[4] R. B. Gozzoli, Hieroglyphic Text Processors, Manuel de Codage, Unicode, and Lexicography,
in: S. Polis, J. Winand (Eds.), Texts, Languages Information Technology in Egyptology,
Presses Universitaires de Liège, Liège, 2013, pp. 89–101.
[5] M.-J. Nederhof, A Revised Encoding Scheme for Hieroglyphic, in: Proceedings of the XIV
Computer-aided Egyptology Round Table, Pisa, Italy, July 2002, IE2002, 2002.
[6] J. Buurman, N. Grimal, M. Hainsworth, J. Hallof, D. van der Plas, Inventaire des signes hiéro-
glyphiques en vues de leur saisie informatique: Manuel de codage des textes hiéroglyphiques
en vue de leur saisie sur ordinateur, Institut de France, 1988.
[7] M.-J. Nederhof, The Manuel de Codage Encoding of Hieroglyphs Impedes Development
of Corpora, in: S. Polis, J. Winand (Eds.), Texts, Languages Information Technology in
Egyptology, Presses Universitaires de Liège, Liège, 2013, pp. 103–110.
[8] M.-J. Nederhof, Automatic Alignment Of Hieroglyphs And Transliteration, Gorgias Press,
2009, pp. 71–92. URL: https://doi.org/10.31826/9781463216269-007. doi:doi:10.31826/
9781463216269- 007 .
[9] D. A. Werning, Towards Guidelines for TEI Encoding of Text Artefacts in Egyptology: TEI
Templates, Thesauri, ”EpiDoc” cheatsheet, 2016. URL: https://nbn-resolving.org/
urn:nbn:de:bsz:15-qucosa-221617.
250