From MASTER to TEI P5

1. Content model changes

  1. Many elements in MASTER had a content model of p+. In TEI P5, all such elements have a content model of macro.specialPara. The difference is that there is no need to introduce a <p> element to wrap the content of the element. For example, a MASTER record such as
    <accMat><p>Pastedown on fols 2 and 4</p></accMat>
    could appear in TEI P5 simply as
    <accMat>Pastedown on fols 2 and 4</accMat>
    Elements with this content model are: <accMat>, <acquisition>, <additions>, <collation>, <condition>, <custEvent>, <decoNote>, <foliation>, <layout>, <musicNotation>, <origin>, <provenance>, <source> and <surrogates. The element <watermarks> also had this content model>, but has been removed (see below).

    Question: when converting from MASTER to P5, should the unnecessary <p> be retained or removed?

    Resolution: The ENRICH project decided that where multiple <p> elements exist, they should be kept, but where only a single <p> element exists in a TEI element where it is optional, then the wrapping <p> should be discarded unless it contains attributes recording important intellectual content.

  2. Many elements in MASTER had a content model in which p+ was alternated with some other, more specific, elements. In most cases the alternation is exclusive -- that is, the element may contain either p+ elements or more specific elements but not a mixture; in a few cases, the alternation is inclusive -- that is, the element may contain a mixture of paragraphs and more specific elements. Elements which are exclusive in this sense are: <adminInfo>, <custodialHist>, <decoDesc>, <history>, <msContents>, <physDesc>, <objectDesc>, and <layoutDesc>. The only inclusive elements are <bindingDesc> and <binding>.

    In TEI P5, the direct references to <p> within all such content models have been revised to reference the element classes model.pLike. This makes it possible in a TEI P5 schema to use either <ab> or <p>, if <ab> is available; or indeed to define other generic elements for use here.

    Question: should <ab> be used?

    Resolution: The ENRICH project did not express a preference for the use of <ab> instead of <p>. They recognised that there was semantic baggage with <p>, they felt it was an unnecessary change.

  3. The following elements, which all had a content model of PCDATA in MASTER, in TEI P5 have a content model of macro.xText: <msName> (formerly <altName>), <collection>, <depth>, <height>, <institution>, <locus>, <origDate>, <origPlace>, <repository>, <width>. This change permits the inclusion of the <g> element used in TEI P5 to represent nonstandard characters or variant glyphs in otherwise straightforward text.

  4. In MASTER, the <msHeading> element was used to provide a brief description or characterization of a manuscript. In P5, this has been replaced by a generic <head> element, which contain most -- though not all -- of the child elements previously allowed within <msHeading>: specifically, it does not permit <author>, <respStmt>, or <textLang>, although a very large number of phrase level elements are permitted. Driscoll points out that most of the components of <msHeading> are better provided elsewhere in the record, and that the heading could simply be untagged text.

    In the Manuscriptorium document on Technical Compatibility of Metadata, however, amongst other rules about minimal compatibility requirements, the recommendation is made that <msHeading> should include as a minimum <title>, <author>, <origdate>, <textLang> and (optionally) <note>. This is not enforced by any DTD however.

    Question: Should the schema enforce the Manuscriptorium recommendations for <head>? if so, should we rename the element?

    Resolution: It was felt by a few members of ENRICH that the need for an existing, possibly structured, brief description was beneficial even if this resulted in a duplication of information. The ENRICH project decided that <head> should not be renamed for the ENRICH schema, and could act as a secondary location to record summary/overview metadata, but that the proper places in the header will be used in preference by processing software. Thus, if a <msHeading> exists which contains unique information, this will be copied to the correct places, and transformed to a valid <head>, but it no such <msHeading> exists, then it will not be created. If information exists in the proper locations and a <head> then the proper locations should always be used when providing the information. The project accepts that <name type="author"> is a good enough replacement for <author> in such a field. However, the project wonders if <lang> could be used instead of the missing <textLang> that is unavailable at this point, or if <textLang> could be added to model.pPart.msdesc and thus be available in macro.paraContent

Other miscellaneous content model differences are listed below:
  1. the <dimensions> element in TEI P5 may contain zero or one <height>, <width>, and <depth> (in that order) rather than any number of such elements in any order.
  2. the element <accMat> becomes a sibling of <additional> rather than a child of it
  3. the element <remarks> within <adminInfo> is replaced by a reference to the class model.notelike
  4. the element <overview> is not permitted within <bindingDesc>: if one is provided, its contents will be converted to <p> elements.
  5. the element <overview> is replaced by the element <summary> when it appears within <msContents>
  6. the element <langUsage> is replaced by <textLang> when it appears within <msItem>
  7. the element <q> is replaced by <quote> when it appears within <msItem>; <cit> (grouping a quote with an attribution) is also possible.
  8. the element <msContents> may contain an optional <textlang> element
  9. a new element <msItemStruct> may appear within <msContents> as an alternative to <msItem>
  10. the <handDesc> (formerly <msWriting>) element now contains an exclusive alternation of model.pLike elements with <handNote> (formerly <handDesc>) elements, rather than an inclusive one.
  11. The <support> has been given a simpler content model (macro.specialPara) and the optional specialist elements (<overview> and <watermarks>) it contained are no longer available within it.

Up: Contents Next: 2. Elements renamed