Introduction‎ > ‎Document Structure‎ > ‎

Narrative Structure

There are many uses for narrative text but the two most important categories are for transcriptions and for generating new narrative work (e.g. essays, reports, inference, etc.). These have markedly different characteristics that STEMMA tries to streamline:


  • Transcription (including transcribed extracts) — requires support for textual anomalies (uncertain characters, marginalia, footnotes, interlinear/intralinear notes), audio anomalies (noises, gestures, pauses), indications of alternative spellings/pronunciation/meanings, indications of different contributors, different styles or emphasis, and semantic mark-up for references to persons, places, groups, animals, events, and dates. The latter semantic mark-up also needs to clearly distinguish objective information (e.g. that a reference is to a person) from subjective information (e.g. a conclusion as to whom that person is).
  • Narrative work — requires support for layout and presentation. Descriptive mark-up captures the content and structure in a way that provides visualisation software with the ultimate control over its rendering  It needs to be able to generate references to known persons, places, and dates that result in a similar mark-up to that for transcriptions. The difference here is that a textual reference is being generated from the ID of a Person entity, say, as opposed to marking an existing textual reference and possibly linking it to a Person with a given ID. Also needs to be capable of generating reference-note citations and general discursive notes.


The basic structure of a narrative block is as follows:




<Narrative Key=’key’>

            [ <Title> narrative-title </Title> ]






<Text [Key=’key’] [TEXT_TYPE] ...  [DATA_ATTRIBUTE] ... >

[ <Title> text-title </Title> ]




<FromText Key=’key’/>


A single <Narrative> entity is divided into separate rich-text <Text> segments, each of which has independent properties controlled by the attributes listed below. When <Text> elements are nested, their stacked properties are merged, and then un-stacked when the nesting finishes. Elements with separate opening and closing tags must be closed before the corresponding </Text>.


The Text segments can reference the Keys of arbitrary main entities using the semantic mark-up defined below.


The <Text> element has an optional Key attribute that allows it to be referenced or utilised from elsewhere. The <FromText> element also allows the content of a named <Text> element to be re-used as though it were physically present in the current narrative container.



<Text Key=’tDemiseJessamine-en’ Language=’eng’>

<Title> Demise of Jessamine Cottages </Title>

<PlaceRef Key=’wJessamine’ Mode=’Hierarchy’/>, were demolished in <DateRef Value=’1956’/>.


<Text Key=’tDemiseJessamine-fr’ Language=’fra’>

<Title> Disparition de Cottages Jessamine</Title>

<PlaceRef Key=’wJessamine’ Mode=’Hierarchy’/>, ont été démolies en <DateRef Value=’1956’/>.




This text could be referenced from another Text section using the appropriate Key name in a <NoteRef> mark-up element. It might generate the following English text on the screen when loaded by an appropriate viewer:


Jessamine Cottages, Nottingham, were demolished in 1956


The nature of the text in a Text segment may be characterised using one-or-more of the following attributes:




Language=’code’ | Locale=’code’

Sets Language or Locale value for the text segment. These are discussed in the section on the overall Document format.


Class=Header | H1 | H2 | H3 | Caption | Legend | Endnote | Footnote | Tablenote

Characterises the content of the text segment; software can apply appropriate styles based on this class. Header marks the section as containing header information for the main body. Although this may be used for transcriptions (e.g. of a letterhead), it is primarily designed for the header containing authorship, title, etc., in a narrative work (including essays and reports). H1H3 are header levels for section headings. Caption may be used to set image (see ResourceRef) and table captions. Footnote, Endnote, and Tablenote are for text to be displayed at the bottom of a page, document, or table. See



Flags a <Text> section as a translation of a cited source from its original language, as specified by the given ISO 639-2 three-letter language code, to the current language of the enclosing <Text> element.







The values for ‘level’ are: Public (default), Family, Private, and Sensitive. Access is granted based on which 'family' the requestor belongs to. Hence, public is OK to everyone, family is OK to an appropriate family member, private OK to ad hoc people selected by the owner, and sensitive to no one but the owner. These values are part of the controlled vocabulary associated with the namespace.



Some percentage of how certain the data is. Default=100%. The surety of a datum or inference is more specific than the confidence in a data source or the information derived from it. See Probabilities.



Indicates the text or datum is inferred from other data. Inference, as used here, implies the reasoning (or proof argument) as well as any conclusion. The default is ‘0’ (i.e. False) implying information rather than reasoning or conclusion. In conjunction with Surety, a value of ‘1’ (i.e. true) can therefore represent conjecture, an educated guess, or even pure speculation.



Part of a details-link that connects conclusions to their explanation, evidence, and original information. See Source for more information.



For instance:

<Text Sensitivity=’Private’ Surety=’20%’>...some sensitive comment that I'm not sure about...</Text>