In this section we detail the characteristics of the text we are editing. An example that has been filled out:

<textDesc n="epic poetry">
	<channel mode="w"/>
	<constitution type="single"/>
	<derivation type="original"/>
	<domain type="art"/>
	<factuality type="fiction"/>
	<interaction type="none"/>
	<preparedness type="prepare"/>
	<purpose type="entertain" degree="high"/>
</textDesc>

The options for the name @n in our classification of the <textDesc> are: "epic poetry", "tragedy", "mythography", "history". We can add other variants as the corpus grows.

The tags contained in <textDesc> do not carry a node, only attributed data, which are predefined and need to be selected manually. Nodes can be added if any clarification is required.

• <channel>: the possible values of the transmission channel are:
  • s (spoken) 
  • w (written) 
  • sw (spoken to be written) e.g. dictation
  • ws (written to be spoken) e.g. a script
  • m (mixed) 
  • x (unknown or inapplicable)
• <constitution>: indicates the state in which the text is preserved:
  • single – a single complete text
  • composite – a text made by combining several smaller items, each individually complete
  • frags – (fragments) a text made by combining several smaller, not necessarily complete, items
  • unknown – composition unknown or unspecified
• <derivation>: nature and degree of originality of the text:
  • original – text is original
  • revision – text is a revision of some other text
  • translation – text is a translation of some other text
  • abridgment – text is an abridged version of some other text
  • plagiarism – text is plagiarized from some other text
  • traditional – text has no obvious source but is one of a number derived from some common ancestor
• <domain> : social context for which the text was composed or created:
  • art – art and entertainment
  • domestic – domestic and private
  • religious – religious and ceremonial
  • business – business and work place
  • education – education
  • govt – (government) government and law
  • public – other forms of public context
• <factuality>: the extent to which the text is fiction or non-fiction:
  • fiction – the text is to be regarded as entirely imaginative
  • fact – the text is to be regarded as entirely informative or factual
  • mixed – the text contains a mixture of fact and fiction
  • inapplicable – the fiction/fact distinction is not regarded as helpful or appropriate to this text
• <purpose>: indicates the purpose or communicative function of the text. Attributes can be free, e.g.:
  • persuade – didactic, advertising, propaganda, etc.
  • express – self expression, confessional, etc.
  • inform – convey information, educate, etc.
  • entertain – amuse, entertain, etc.

All this data is relevant when classifying documents, and will vary slightly from one work to another, so the editor should complete it carefully.