Markup vs. Character Encoding: The quandary of handling the epigraphical/papyrological "underdot" in computer representation

poster / demo / art installation
  1. 1. Deborah Winthrop Anderson

    Department of Linguistics - University of California Berkeley

Work text
This plain text was ingested for the purpose of full-text search, not to preserve original formatting or readability. For the most complete copy, refer to the original conference program.

Markup vs. Character Encoding: The quandary of handling
the epigraphical/papyrological “underdot” in computer representation

Dept. of Linguistics, UC Berkeley


New York University

New York, NY





When dealing with ancient texts written on various media for detailed
scholarly publication, it is critical to convey information on the specifics
of the writing. After a photo, scanned image, or line drawing is made from
the original text, texts are commonly transferred to paper or an electronic
medium. In order to capture the information from the inscription,
transliteration and transcription schemes in Roman letters (or Greek, for
materials using Greek script) are often used to capture all the
characters--whether clearly legible, faint, damaged--and the empty
Ancient texts, especially those in damaged condition, present difficulties,
for they must rely upon the subjective judgment of the transcriber (and
editor) in deciding what characters are present, whether an erasure is
identifiable, the amount of empty space(s), etc. Certain conventions on how
to represent these details have been created and are followed in various
fields (i.e., double square brackets [[ ]] enclose erasures, angle brackets
< > indicate letters made in err by the scribe). A common method in
transliteration and transcription for denoting a damaged character--or one
whose identity is uncertain--is a dot placed below a letter. This underdot
is common in ancient Greek texts and Latin, for example, at least those
editions intended for the scholar interested in paleography, philology,
A problem arises in how to handle the underdot in computer representation.
This question surfaced in a project done at UC Berkeley in conjunction with
the Berkeley Library, wherein the Indo-European Studies
Bulletin, a publication affiliated with the UCLA Indo-European
Studies Program, was being put online, using XML, Unicode, and a TEI-Lite
DTD. The underdot appeared in a Sabellian inscription. Since our project
intended to test out the use of Unicode, we reviewed the options available
in Unicode and employed the combining underdot (U+0323) after the character.
On the surface level, this reflected the character represented in the
Sabellian article. However, the combining underdot raised a potential
problem: by interrupting the plain text string with the diacritic, searching
for the entire word could be impeded, unless the underdot was taken into
account when searching. More importantly, should the underdot actually be
encoded as a separate character? Is it on the same level as an “a acute”,
for example, where the diacritic is an essential part of the character? Or
should markup be used, such as denoting the sign with a <damage>
and/or <unclear> tag? Markup then could be visually rendered according
to one’s own convention or taste.
Since Unicode support is only just now becoming more prevalent in new
software and hardware, most computer projects have adopted ASCII
representations of the underdot and other epigraphic and papyrological
symbols. For Greek, the Beta Code of the Thesaurus Linguae Graecae has been
widely adopted by projects (such as Perseus). Eventually, a changeover to
Unicode will occur, and the need to decide how to handle it is becoming more
pressing: should one use a character encoding or markup?
The underdot is but one member of a long list of epigraphic and papyrological
symbols used in transcription and transliteration. An agreement amongst
scholars ought to be made if there is to be consistency in handling these
symbols within the same discipline and across disciplines, since similar
problems are faced in other fields. Currently, Unicode proposals on
cuneiform, Coptic, and Iranian await to see how the problem is resolved in
the Greek and Latin sphere, since it will influence their projects. Or are
the variations between fields (and lack of communication so great) that a
discipline-specific approach will prevail?

A number of important issues arise when reviewing the problem more deeply:
--While the underdot is frequently used to indicate damage or
uncertainty, it is not necessarily consistently used with this broad
definition, even in ancient Greek materials. In a standard book used
for Greek dialects, Carl Darling Buck’s The Greek
Dialects (Chicago and London, 1955), he states: “The
occasional use of a dot under a letter indicates that it is
mutilated. But this is commonly disregarded if the proper reading is
reasonably certain” (p. 184). In Mycenaean materials (Emmett
Bennett, Jr., and Jean-Pierre Olivier, The Pylos
Tablets Transcribed, Part I: Texts and Notes, Rome,
1973), however, an underdot under a digit can indicate that there is
a question whether the number is in the text at all, a problem
regarding the identity of the number, or it may merely indicate that
the number is almost illegible” (p. 10). Indeed, if fine
granularity of a text is intended, “damage” and
“uncertainty” can and probably should be separated as two
distinct elements, and this is so done in a new proposal,
“Epidoc”, being worked by at the University of North Carolina by
Tom Elliott, Hugh Cayless, and Helen Hawkins (, ).
--In some languages, an underdot has a specific phonetic meaning.
In Sanskrit, it is used for a retroflex s. The phonetic sense
specific to the underdot is at variance with the unclear sign
meaning. Potential confusion with the phonetic sign is possible in
--One potential problem of character-encoding with Unicode is an
apparent ambiguity of certain signs for the naive user. A scholar
looking for a combining underdot when skimming through the Unicode
Standard (or scrolling down the choices under MS Office 2000’s
Arial Unicode MS font) may choose, quite incorrectly, U+093C, the
Devanagari sign nukta, which has very specific use. This error would
cause problems for searching and rendering.
--A number of characters for damaged signs are proposed in a
Unicode proposal for Egyptian hieroglyphs. While such signs would
appear with the hieroglyphic characters and not in a Roman-type
transliteration/transcription scheme, it significantly offers a
character-encoded model for conveying a damaged sign, and not one
based on markup. Could this option peacefully co-exist with a
markup-only approach used in other projects and is this
--Unicode will allow using the ancient scripts more fully, since
the character encoding standard should allow for easier writing,
rendering, and printing of the original scripts, beyond what printed
publications have been able to offer in the past. (However, this is
only possible with necessary Unicode-enabled operating system,
software, and font support for the characters are present.) Hence, a
fuller representation of a text with the ancient scripts could be
added between the layers of photo/drawing and Romanized
transliteration/transcription. Instead of using an underdot to
indicate a faint letter, for example, markup could be used with the
original script (as well as the transliterated/transcribed version)
to make the sign (or letter) appear fainter or in a slightly
different color.
--A consistent markup scheme could be used with a style sheet to
render the faint/damaged letter in a variety of ways, as suggested
above, offering wide extensibility. If a text is intended for
beginners, the markup indicating the traces of letters or erasures
could be disregarded. If Hittite scholars regularly used a special
symbol for a mutilated sign, for example, this could also be
accommodated by changes to the style sheet.
--Since new Unicode character encoding proposals can take from two
to five years from the first proposal until approval into ISO 10646,
markup offers a much quicker solution. A “best practices” guide
to markup, similar to the Epidoc proposal, would be needed.

A Possible Solution?
Markup seems to present the best option, for it allows flexibility and
provides a speedier means to put epigraphic/papyrological text information
on the Web. Also, the use of the underdot reflects information regarding
damage/uncertainty/etc. of a character, and hence is probably best not
encoded as a separate character. However, if markup is to be advocated as
the best approach here, can user-friendly software be created in the
foreseeable future for typing and rendering of such markup schemes? This
poster is intended to encourage further open discussion on the “underdot
quandary” and its ramifications, and to seek input from others,
particularly those with projects on ancient texts or whose expertise is on
markup and relevant technology.

If this content appears in violation of your intellectual property rights, or you see errors or omissions, please reach out to Scott B. Weingart to discuss removing or amending the materials.

Conference Info

In review


Hosted at New York University

New York, NY, United States

July 13, 2001 - July 16, 2001

94 works by 167 authors indexed

Series: ACH/ICCH (21), ALLC/EADH (28), ACH/ALLC (13)

Organizers: ACH, ALLC