The Theological Markup Language is defined as a set of extensions to the HTML 4 Strict DTD. The DTD for XML is available at http://ccel.wheaton.edu/ThML/dtd/ThML093.dtd. Since the ThML DTD includes HTML 4, it is technically an SGML language and not an XML language. However, it is intended that the extension to HTML are compatible with XML. It may be possible to validate a ThML document with and SGML parser and use it as well-formed XML.

The ThML.body element contains all of the contents of the print edition of the book on which the text is based. The ThML.head element contains bibliographic and meta information about the text, both the electronic publication and the print edition upon which it is based, if any. The information in this section is not generally a part of the original book but taken from MARC records or added to document the publication of the electronic version. It may also include keywords, information about the program used to generate the text, etc.

Structural divisions in the body of the text are marked with <divn> tags as in this example:

<div1> is used for top level parts of a text, including, for example, a title page, preface, table of contents, chapters, index, etc. Additional levels are used for lesser structural divisions of the document. These structural divisions show the structure of the original text, and they are also used to prepare a table of contents and allow splitting of a text info files and access to a text by section. The optional title attribute is used in constructing a table of contents and may be used for running heads or other identification purposes. The optional type and n attributes may be used to specify the type and number of the division. If they are present, they will also be used to identify the section in the table of contents.

The <insertContents level="3" /> tag may be used to specify that a table of contents be inserted at that location. In the example above, the title attribute of the <div1> tag would be used as the title of the Table of Contents, and all of the all of the <div1> and <div2> entries would be gathered and listed hierarchically. Each entry would be linked to the appropriate section.

It is often useful to know the page breaks from the print edition of a book. They may be used as targets for subject index entries identified by page number or to display a text with the pagination of the print edition. Page breaks are marked by the insertion of <pb /> tags, with the n attribute giving the page number of the upcoming page (<pb n="37" /> or <pb n="xii"/>). These elements should appear at the start of the identified page. Many electronic texts will also have images of pages available on line. The pb element will also take an href attribute specifying a URI for an image of the page (<pb n="37" href="gif/0021a.gif" />).

Scripture references may occur in footnotes, in parentheses (Phil. 2:1-11), or in the text itself -- see Rom. 8:28. Context may be needed in order to interpret a reference -- see verse 29 and 10:8-13. Several passages may be stacked together in one citation (Matt. 5:44, 46; Luke 7:42; John 5:42, 13:35, 14:15, 23; 15:12-13; 21:15-16; 3 John 13). For marking scripture citations, ThML will use the <scripRef> element, as in this example:

Also, context is sometimes necessary in interpreting a reference. A passage may refer to Romans 8:28 at one point and later to verses 29 and 30 and chapter 10:8-13. A parser should be able to identify the context in most cases, but in some cases it may be necessary to set the context or turn the parser off. The <scripContext version="NIV" passage="Romans 8" /> element is used to set the default context for the parser, and the <scripParseOff /> and <scripParseOn /> elements may be used to turn the parser off or on, to prevent linking of a passage such as "Bob had 2 apples and John 3." The version attribute may be set in a scripContext element but it is never set by the parser.

In theological texts, scripture is also sometimes quoted. In this case, it is not desirable to link the reference to the scripture passage, but it may be desirable to incorporate the passage into a table of scripture references. Quotations of scripture may be marked with the <scripture> element. A passage may be represented as in this example:

Explanation or commentary on a passage involves a semantic relationship between the explanation and the passage explained. This relationship should be represented in the text in order to be able to build an index of scripture commentary. For example, it would be useful to be able to see everything the early church fathers, said or preached on a passage. Commentary or explanation of a passage will marked with a <scripCom> element, as in this example:

Standardized symbolic cross reference schemes such as dates, keywords, or Strongs numbers aren't really links to other documents, because any two documents with compatible cross reference schemes can be linked together and no particular documents are intended. Therefore XLL links, element IDs, etc. don't capture the semantics of such information. We will define a new sync element to represent this information. For example, the element

The scheme name given in the type attribute are not pre-defined; applications may invent new synchronization types for specific purposes. For example, the Rule of Benedict is available in several different manuscripts in two different traditions. If a common synchronization scheme were defined and manuscripts marked up, any two or more could be selected and aligned as parallel columns on the screen, or alternate forms of a passage could be located.

In a ThML document, footnotes, endnotes, etc. are all marked with the <note> tag, following the syntax used by TEI Lite for the most part. The note element may take the following attributes: place, resp, target, targetEnd, and anchored. The place attribute specifies how it appears in the text (e.g. end, foot, inline, or margin). The target (and targetEnd) attributes refer to the start (and end) of the text being annotated, if the note does not occur in the text at its reference point. These attributes allow the notes to be gathered at the end of a chapter or file if desired. The resp attribute identifies the person responsible for the note -- for example, the author, editor, or a person's initials. The anchored attribute make take the value yes (default) or no, specifying whether the note is anchored at an exact location; margin notes typically are not anchored.

The <note> element can also be used to store commentary, margin scrawls, and the like for a text in a separate file. In that case, the target and targetEnd attributes would be references to a point in another document.

The primary language for a document is specified in the header. Passages in other languages may be marked with the foreign tag and the lang attribute. For example, the Greek passage <foreign lang="el">logos</foreign> may be marked as shown. "lang" attribute values are as specified in ISO 639. Some examples are Dutch: nl, English: en, French: fr, German: de, Greek: el, Hebrew: he, Latin: la, Spanish: es, Portuguese: pt, Russian: ru. Note that the lang attribute may be used with most elements.

If the language uses characters not available in the ISO-8859-1 (Latin-1) character set, they may be represented in Unicode (preferred), as in this Greek example (λογος) and this Hebrew example (הלהי), or using the Latin-1 character set and a suitable font, for example, <foreign lang="el" style="Font-family: SIL Galatia">logov</foreign>. The Greek and Hebrew fonts recommended for use with the CCEL are the freeware SIL Galatia and SIL Ezra fonts and related software from the Summer Institute of Linguistics, used here in a Greek example (logov) and a Hebrew example (hwhy).

Theological books often contain verse -- poetry, hymns, or versified presentation of material such as the Psalms. A stanza, verse, or other unit of verse is encoded in a <verse> element. Verse is often written with varying levels of line indentation. Lines are marked with <l>, <l2>, and <l3> elements, identifying relative levels of indentation. In the example below, the indentation is of course ignored by the XML parser, but it should be reproduced by the presentation software based on the <l>, <l2>, and <l3> tags.

<attr><name>G. Ter Steegen</name></attr>

Attributions of authors of poetry, letters, etc. may be marked with the <attr> element. This might be rendered right-justified and italicized. Also, names may be marked with the <name> element. When they are thus marked, an index of names can be automatically constructed. A different representation of the name for the index may be specified with the title attribute: <name title="Ter Steegen, Gerhard">G. Ter Steegen</name>. (The title attribute may be used with any element; it is used for index entries and tables. In web-based presentation of a book, it may also be used as a "tool-tip".)

Citations of other works such as books or treatises may be marked with the <citation> element. That element may also take an href attribute to specify a URI for the cited work, if available. The <date> element may be used to mark dates that occur in the text. A value attribute may be used to specify the date in ISO format, as in this example: <date value="1997.12.25">last Christmas</date>. The insertIndex element (described in the next section) may be used to insert an index of works cited, dates, or personal names. In each of these cases, the intent is to assist searching and indexing. Searches for names can be restricted to names only, and an index of dates mentioned in a book can be constructed.

Passages in the text may be marked for insertion into an index using the <index> element. For example, one might mark a passage for inclusion in a subject index this way:

The title attribute is used as the identifier in the Table of Contents. If it is not present, the text inside the <index> element is used as a title. The type attribute is used to identify the index that this reference is to be added to; legal values include subject (the subject index for the book) and globalSubject (the library-wide subject index). Other values may be used for specialized indexes.

A document may use several user-selected types of index entries. An XML element (<insertIndex type="subject" />) is also provided to specify that a sorted, hierarchical index of all the "subject" (e.g.) index entries should be inserted at that point, with links to the appropriate locations in the text. Certain additional index types are also understood: <insertIndex type="name" /> inserts an index of all names marked with the <name> element; if the title attribute is present, it is used as the index entry. Similarly, indexes may be inserted for citations, dates, foreign words and phrases, images (<img>), names, scripture references (<scripRef>), and scripture commentary (<scripCom>).

Some documents contain a glossary. Glossaries may be marked up with the <glossary> element and HTML <dl> (definition list), <dt> (term) and <dd> (definition) elements. The glossary may take an optional type attribute, to specify how the glossary may be used.

The <body> section of the electronic text should have all of the contents of the print edition. However, for display purposes, it may be desirable to add or delete to the print edition. For example, it may be desirable to delete the original table of contents and replace it with one that is automatically generated. The <added> element is used to mark sections that have been added and do not appear in the print edition, and the <deleted> element is used to mark the sections that have been deleted and should not appear when the book is presented, even though they remain in the XML for those who want to see exactly how the print edition looked. For example, a table of contents might be marked this way:

The head section of a ThML etext has the most detailed (and least frequently used) markup. In a practical ThML software system, much of this information will be filled in by making entries in a form or template, including pasting the MARC record for the print source into the form. The head section may start with some HTML elements, such as <title> and <meta>. In addition, it has three optional sections that are unique to ThML: <generalInfo>, <electronicEdInfo> and <printSourceInfo>.

The <generalInfo> section contains information about the text that is not specific to the electronic edition or the print edition on which it is based. Whenever possible, its components are filled in from the information in the MARC record. It may contain these elements: