HMT editor's guide: guide to XML usage for editors

Introduction

This document describes the HMT project’s use of TEI XML in the summer of 2017.

All citable texts in the HMT project follow the OHCO2 model: citable units are organized in an ordered hierarchy of citation units. We use familiar TEI elements to represent the citation hierarchy: citable nodes are either l for lines of poetic text, or p for paragraph units of prose, with higher parts of the hierarchy (such as books of the Iliad) represented by TEI div elements. Each element corresponding to a level of the citation hierarchy has an @n attribute giving identifying value for that part of the text. Iliad 10.1, for example, would be contained in a div element with @n attribute of 10, and a l element with a value of 1.

If the text of a scholion explicitly organizes material in a table or list structure, we use TEI list with item elements. These are the only TEI elements we need to capture the basic structure of our diplomatic editions.

Example: see an example of a list structure in a scholion. Items can be numbered (if they are numbered in the text) by adding the attribute @type="ordered" to the list element and the item elements can take an @n attribute to indicate the sequence.

The focus of this document is our usage of TEI markup within those citable nodes, that is, how we document the contents of our citable nodes. The order of sections describes our markup practice from “inside to out”: when elements co-occur, the element described first in this document (and so, innermost) will be wrapped by the element described later (outermost). The major sections are:

  1. transcription of the main text
  2. tokenization the main text
  3. semantic disambiguation of named entities and abbreviations
  4. scribal intervention in the text
  5. editorial identification of the voice of the text

All editorial activity at every level involves scholarly judgments, and the clean separation suggested by this scheme is to some degree artificial: nevertheless, we hope this schema will be helpful in guiding editors to follow the HMT project’s editorial principles as they exercise their best judgment.

1: Diplomatic transcription

The first task of HMT editors is to transcribe the text as they read it on the manuscript. Editors must understand what belongs in a text edition, what character set to use in transcribing the text, and how to encode the paleographic status of a reading.

What doesn’t belong in a text edition?

Text editions of the Iliad or scholia texts should not contain anything in the following categories:

  1. critical signs: located next to the Iliadic lines in the Venetus A
  2. connecting signs: sometimes written next to scholia and corresponding main text
  3. metacomments: numerical or other symbolic notation that further explains the text (e.g. reordered lines)

Orthography and digital character set

Texts should be transcribed using a specified subset of the Unicode character set.

accentuation and breathings
We record all accents and breathings as given in the manuscript (not “corrected” to modern usage). This category includes diaeresis, and markings for short or long vowel quantities (breve and macron).
iota subscripts
Iotas are consistently written adscript in most of our manuscripts: we write them subscript where that is usual in modern practice without implying that the manuscript does so.

Example: καλῇ (urn:cts:greekLit:tlg0012.tlg001.msA:1.604)

ligatures and variant letter forms
Our transcription does not note ligatures or variant letter forms: readers interested in these visual features of the manuscript should consult the associated high-resolution images. We treat regular abbreviations for morphological endings, for prepositions, and for the conjunction καί as variant letter forms, and silently expand these in the transcriptions. See more on variant letter forms under abbreviations.

Diacritical Marks

Most Greek keyboards are not able to handle more than two diacritical marks, so in instances where you need and accent, breathing, and a diacritical mark, use combining characters which are typically found in special character menus in most basic text editors.

diaereses on vowels with breathing and/or accent should use Unicode “Combining Diaeresis” character.

Example: ί̈σχειν (urn:cts:greekLit:tlg0012.tlg001.msA:9.352)

macra should use Unicode “Combining Macron” character.

Example: δαίνῡ {urn:cts:greekLit:tlg0012.tlg001.msA:9.70)

brevia should use Unicode “Combining Breve” character

urn:cite:hmt:edguide:brevefloat

elisions

Of the many possible Unicode characters for elision, use only Unicode straight single quote. If you use Sophokeys to enter Greek, note that this is not the character it generates automatically when entering Greek.

Example: ἄλλ’ ἐν (urn:cts:greekLit:tlg0012.tlg001.msA:9.235)

“floating” grave accents

Floating grave accents in the manuscript can appear on an elided syllable. “Float” a grave accent character between the last letter of the word and the elision mark.

Example: μετα δ`᾽ ἔσσεται (urn:cts:greekLit:tlg0012.tlg001.msA:9.131)

** Punctuation Used **

When editing manuscripts, it is important to remember that punctuation is an important part of producing a diplomatic edition. Some easy places to forget about punctuation and other similar graphemes include: the end of Iliadic lines, the crosses at the beginning of lemmata , the various forms of punctuation or lack thereof at the end of lemmata, and punctuation at the end of scholia.

period: . high stop: · comma: , colon: : end of scholion: ⁑ cross: ‡

You should only find end-of-scholion markers and the cross in scholia.

Paleographic status

While readers always take many kinds of contextual information into consideration when reading a text, HMT editors should distinguish three levels of legibility based solely on the paleographic clarity of the reading.

  1. clear. The letter is unambiguously legible, based on paleographic considerations alone. It may be incompletely preserved, but the visible remains cannot be read as any other character.
  2. unclear. Part of the letter is visible, but taken by itself cannot be unambiguously read. That is paleographically ambiguous. You might be able to make an educated guess, but that is not a diplomatic edition.
  3. missing. No trace of the letter remains, but it is clear from the context that one or more letters were originally present. This occurs in instances where there is damage to the manuscript such as a hole or extreme fading, even in specialized lighting.

Clear readings

Enter the text with no additional markup

Unclear readings

Use TEI unclear element. If the unclear element breaks up a lexical token or numeric token, wrap the entire word in a TEI w element or num: see the following section on “Tokenization” for more details. Note that w is only necessary if the word is broken up. If a whole word is unclear, w is not necessary and need not be included.

Example:

<w>γ<unclear>ὰ</unclear>ρ</w>

(urn:cts:greekLit:tlg5026.msA.hmt:8.132)

Missing Use TEI gap element. Use @unit attribute for the type of the gap (for example, “letters” or “words”), and @extent to indicate how many units are likely included in the gap.

Example:

<gap unit="letters" extent="3"/>

Scribal modifications

(For markup of more extensive scribal interventions in the text, see section 4.)

2: Tokenization

Editors must subdivide their transcribed text into tokens, and explicitly or implicitly classify each token as one of the following types:

  • lexical type (a “word”)
  • numeric type (normally written with alphabetic characters in Milesian notation)
  • punctuation
  • unintelligible
  • literal string of characters

While print editions typically imply tokenization of all types by separating them with white space or punctuation characters, the use of space and punctuation in Byzantine manuscripts is not a rigidly consistent guide to tokenization, so HMT editors must deliberately distinguish tokens in their transcribed text. For automated processing, punctuation tokens can be distinguished by their unique character set, and so require no markup. Lexical tokens are the default type, and require no further markup if a token is unambiguously bounded by white space or punctuation. Other tokens always require markup, as explained next.

Lexical tokens

If a lexical token is paleographically clear, editors should use white space to separate it from adjacent lexical tokens. White space is not necessary if the adjacent token is a punctuation charater.

If parts of a lexical token are unclear or include gaps, however, XML processors are free to separate the literal text of the clear readings from text included in There are a few cases where we cannot rely on whitespace to split our tokens because of other TEI mark up.

paleographic clarity

As previously noted, transcribing a text can sometimes split lexical tokens with gap or unclear. In these situations, the whole word needs to be explicitly wrapped in the w (“word”) element so that the contents can be recognized as a single token.

scribal additions

Use TEI add element, including @place attribute with value supralinear

Example:

<w>ἔνι<add place="supralinear">οι</add></w>

(urn:cts:greekLit:tlg5026.msA.hmt:4.9)

In cases where the addition does not split a word, you would not need to use w

Numbers

Use TEI num with @value attribute. Numbers are often denoted with a horizontal bar by the scribe to indicate that they are not lexical content.

<num value="50">ν</num>

(urn:cts:greekLit:tlg5026.msA.hmt:1.42)

note: Remember that Dindorf and Erbse will typically take a Milesian numeral and write the full Greek word, but we want a fully diplomatic edition. Also remember that book numbers are treated as titles, not numbers in TEI.

Quoted literal strings

Quoted literal strings are defined as strings of letters that should not be considered parseable Greek words. One example might be the scholion talking about the use of the letter sigma. We do not want the machine to treat a single sigma as a Greek word so use TEI rs element with @type attribute = waw (“word-as-word”). Quoted strings are often, but not always, made visually distinct via a horizontal bar. They can be difficult to distinguish from the horizontal bars that denote numbers and you have to determine from context. The neuter article, prepositions, and other vocabulary related to writing will be your clue.

Example:

<rs type="waw">ν</rs>

(urn:cts:greekLit:tlg5026.msA.hmt:1.1498)

Scribal Deletions

Use TEI del when the scribe has either crossed out content, erased, or marked it with “deletion dots”

Example:

<del>οὐκ εἰς το πρεσβεύειν ἀλλ εἰς τὸ πρεσβεύειν Αἴαντας καὶ Ὀδυσσέα Φοίνικος προέληλυθότος</del> (urn:cts:greekLit:tlg5026.msA.hmt:9.193)

Scribal Corrections

Use TEI sic/corr pair for corrections when the wants to correct a reading that is unintelligible in the text; group the pair in a TEI choice element.

Example: accent corrected by the original scribe

<choice><sic>προσηῦδα</sic><corr>προσηύδα</corr></choice>

(urn:cts:greekLit:tlg0012.tlg001.msA:4.337)

Please note that this mark up is only used by corrections we believe were made by the scribe. That means you cannot offer your own corrections and the scribe could be wrong by our orthographic standards. This is a diplomatic editions and we do not want to make our own judges on grammatical correctness at this stage of editing.

Scribal Multiforms

Use the TEI orig/reg pair to identify the reading in the main text and the alternative reading; group them with TEI choice. The scribe sometimes writes only the letters that are to be changed to create the alternate reading. We put the fully expanded word in the reg element.

Example:

<choice><orig>μεθειέμεν</orig><reg>μεθιέμεν</reg></choice>

(urn:cts:greekLit:tlg0012.tlg001.msA:4.351)

Alternative readings are typically strings of characters or even single letters. You will use the whole word like in the example above.

Abbreviations

Abbreviations without mark up

Regular terminating syllables

Regular terminating syllables (often morphological endings, but not always appearing at the end of a word) are treated as a variant letter form, and are not specially marked in any way. See the paleography guide if you are unfamiliar with the common abbreviations. If accents or other diacritical marks should be present on the abbreviated syllable(s), they can be supplied in the expansion.

Example:

εἰκος, where a raised omicron is a brief writing for the morphological ending -ος (urn:cts:greekLit:tlg5026.msA.hmt:4.478)

Symbol Abbreviations

Regular abbreviated forms of the whole words (e.g. καί, ὅτι, δέ, γάρ, διά, etc.) are treated as variant letter forms, and are not specially marked in any way. The rule of thumb for determining if it is a symbol abbreviation is if the symbol cannot be completely broken down into distinguishable letter forms. If accents or other diacritical marks should be present on the abbreviated word, they can be supplied in the expansion.

Examples:

(urn:cts:greekLit:tlg5026.msA.hmt:4.437)

(urn:cts:greekLit:tlg5026.msA.hmt:4.8)

Abbreviations that Require Mark up

paleographically ambiguous abbreviations

Unlike the examples above, the marks that denote these abbreviations do not always indicate the same missing letters. Therefore, because they are paleographically ambiguous, they require mark up. If accents or other diacritical marks should be present on the abbreviated syllable(s), they can be supplied in the expansion.

  • Use the TEI abbr/expan pair, wrapped in choice

Example:

<choice><abbr>ουτ</abbr><expan>οὑτῶς</expan></choice>

(urn:cts:greekLit:tlg5026.msA.hmt:4.495)

Example:

<choice><abbr>περ</abbr><expan>πάτερ</expan></choice>

(urn:cts:greekLit:tlg0012.tlg001.msA:4.288)

3: Semantic disambiguation

Tier 3 semantically disambiguates tokens. This class of mark up is primarily named entities.

Named entities (proper nouns and adjectives)

personal names

For people, mythological or historical, use TEI persName element; include an @n attribute with the full URN value from the [reference table of identifiers for personal names][pers]. If a person does not yet exist in the authority list, create an issue in the repository to request its addition.

Example:

<persName n="urn:cite:hmt:pers.pers1">Ἀχιλλεύς</persName>

(urn:cts:greekLit:tlg0012.tlg001.msA:1.58)

  • Note in cases where more than one person is refered to, such as the Atreidai, you double wrap personal names

Example:

<persName n="urn:cite:hmt:pers.pers22"><persName n="urn:cite:hmt:pers.pers119">Ἀτρεΐδαι</persName></persName>

(urn:cts:greekLit:tlg0012.tlg001.msA:1.17)

place names

For places, mythological or historical, use TEI placeName element; include on the @n attribute has a full URN value from the [reference table of identifiers for place names][place]. If a place does not yet exist in the authority list, create an issue in the repository to request its addition.

Example:

<placeName n="urn:cite:hmt:place.place114">Ἄργεϊ</placeName>

(urn:cts:greekLit:tlg0012.tlg001.msA:1.30)

ethnic groups

These are exclusively for groups of people that can be traced either to a geographic place or an emponymous ancestor. Use TEI rs element. Include a @type attribute with value ethnic, and @n attribute with an identifier from the [reference table for place names][place] or the [reference table for personal names][pers], when you need to use an eponymous ancestor (such as the Danaans).

Example:

<rs type="ethnic" n="urn:cite:hmt:place.place96">Ἀχαιούς</rs>

(urn:cts:greekLit:tlg0012.tlg001.msA:1.61)

  • Since some ethnic names could be attributed to a location or an eponymous ancester (e.g. Trojans link to Tros or Troy), always take the geographic location before resorting to an eponymous ancestor.
  • Note that the ethnic mark up is only for people or groups of people (e.g. Trojan man or Trojans). We would not use it to describe a dialect (e.g. Ionic) or an inanimate object. Those are treated as normal vocabulary items that can be searched.
  • Groups that cannot be tied to a geographic place or an eponymous ancestor are also treated as vocabulary items (e.g. Centaurs, Myrmidons, Gorgons, Giants, Fates, etc).
  • Adjectival forms of people’s names are also not tagged (ergo, we can tag ‘Homer’ but not ‘Homeric’)
  • We also do not tag unclear epithets, even if the identity can be determined from context. For example “Phoebus Apollo” is ok, but the “Earthshaker” to refer to Poseidon is not. If you are in doubt, refer to the description of the indiviual in the authority list. If the epithet is not listed there, you can request it, via Issue Tracker and our editorial guide team can render a judgement.
astronomical bodies

Use TEI rs element. Include a @type attribute with value astro, and @n attribute with an identifer from the reference table for astronomical bodies.

Example:

<rs type="astro" n="urn:cite:hmt:astro.1">Ὠρίωνος</rs>

(urn:cts:greekLit:tlg0012.tlg001.msA:18.486)

4: More extensive scribal intervention in the text

Scribal Deletions

Use TEI del when the scribe has either crossed out content, erased, or marked it with “deletion dots”

Example:

<del>οὐκ εἰς το πρεσβεύειν ἀλλ εἰς τὸ πρεσβεύειν Αἴαντας καὶ Ὀδυσσέα Φοίνικος προέληλυθότος</del> (urn:cts:greekLit:tlg5026.msA.hmt:9.193)

Scribal Corrections

Use TEI sic/corr pair for corrections when the wants to correct a reading that is unintelligible in the text; group the pair in a TEI choice element.

Example: accent corrected by the original scribe

<choice><sic>προσηῦδα</sic><corr>προσηύδα</corr></choice>

(urn:cts:greekLit:tlg0012.tlg001.msA:4.337)

Please note that this mark up is only used by corrections we believe were made by the scribe. That means you cannot offer your own corrections and the scribe could be wrong by our orthographic standards. This is a diplomatic editions and we do not want to make our own judges on grammatical correctness at this stage of editing.

Scribal Multiforms

Use the TEI orig/reg pair to identify the reading in the main text and the alternative reading; group them with TEI choice. The scribe sometimes writes only the letters that are to be changed to create the alternate reading. We put the fully expanded word in the reg element.

Example:

<choice><orig>μεθειέμεν</orig><reg>μεθιέμεν</reg></choice>

(urn:cts:greekLit:tlg0012.tlg001.msA:4.351)

Alternative readings are typically strings of characters or even single letters. You will use the whole word like in the example above.

5: Discourse disambiguation

Quotation and citation

unidentified quoted words, phrases, or passages

Use TEI q for quoted words, phrases, or passages. A quoted selection must be at least one full parsable Greek work. Quoted phrases can ranges from hypothetical phrases to multiforms to quotations from works that we cannot cite. The scribe will not make quotes visually distinct. Use our print editions to help you, as well as context.

Example:

<q>πρὸς ὃ</q>

(urn:cts:greekLit:tlg5026.msA.hmt:1.1498)

identifiable quotations

Use TEI ref/q pair wrapped in a cit element. On ref element use @type attribute =urn. Inside ref element give the full cts urn for the quote (i.e. the text reference). The most common references will likely be Iliad or Odyssey quotes. If you have other citations, see the lists of citable authors and works. If you have a known text that is not referenced, create an issue in the repository. The scribe will not make quotes visually distinct. Use our print editions to help you, as well as context.

Example:

<cit><ref type="urn">urn:cts:greekLit:tlg0012.tlg001:17.453</ref><q>ἔτι γάρ σφισι κῦδος ὀρέξω</q></cit>

(urn:cts:greekLit:tlg5026.msA.hmt:4.8)

Other Discourse Analysis

titles

If the title refers to a known, citable work, use TEI ref with @type='urn' and @n attribute with a full URN for the work

  • Use TEI title element (wrapping ref if that is given). ref is not necessary if it is a work or portion of a work we do not have a URN for. You can find a full list of our citable authors and works. If a known work is not present, please create an issue in the repository to request its addition.

Example:

<title><ref type="urn" n="urn:cts:greekLit:tlg0012.tlg002:">Οδυσσεία</ref></title>

(urn:cts:greekLit:tlg5026.msA.hmt:1.4)

  • Note that book numbers are considered titles, as are common names for books or portions of books (e.g. the Catalog referring to the Catalog of Ships in Book 2).

Titles that are not citable with a cts URN can still be cited with an object URN found at citable works.

Example:

<title n="urn:cite2:hmt:citedworks:work2">Κυπρία</title>

Like the named entities, new citable works can be proposed using the Issue Tracker on the authority list repository.


Website © 2019-2020, the Homer Multitext project. For licensing on image collections, see the Image Archive page.

Powered by Hydejack v8.1.1