Customization of TEI schema for NKJP (text level), version 0.6.4Piotr Bański and Adam Przepiórkowski
This template file is not freely available until it gets released in the (pre)final form.
Created from scratch, based loosely on Sebastian Rahtz's "TEI for Corpora"
NKJP Corpus -- schema for the text layer
This is the schema for the text layer of the NKJP corpus (http://www.nkjp.pl/).
Base: "TEI for Corpora". Removed
typeNote from the header.Removed the "corpus" and "namesdates" modules.Removed numerous elements from the included modules, to
leave as few options as possible. Some elements were retained
for the sake of the header.Defined new content for
body, div, and ab, in terms of direct
references. body can now contain only a mixture of
(div and gap) OR (ab and
gap). div can only contain a mixture of
ab and gap. ab can only contain text,
gap or hi (which is present in legacy
texts).Restricted the root to teiCorpus.Modified the header: category can now contain a
sequence of catDesc, rather than just a single
instance of it.Added modules "corpus" and "namesdates", for
particDesc and listPerson.AP: Added nkjp:topic to the header.AP: Added the nkjp:subcorpus attribute.AP: Added the "spoken" module for recordingStmt.AP: Changed the modality ("usage") of the attribute nkjp:file (actually added earlier) in the header element extent to "req".AP: Reintroduced u within body, for spoken data.AP: Reintroduced sequence of p within div and within body.AP: General clean-up, changing order of element specifications so that ediff with other levels easier.AP: Removed (commented out) the modification of category; desc will be used instead of catDesc for the bilingual description of categories.AP: Added module tagdocs, so that gi, att, etc., may be used in headers.AP: Added xml:lang to nkjp:topic.AP: Modified the reportoire of possible values of nkjp:subcorpus.AP: Introduced nkjp:fsLib which may contain fLib and fvLib. Solution adapted from tei_jos.xml (Tomaž Erjavec's "JOS 100k A" corpus), earlier used in NKJP_structure.xml.AP: In connection with the above, introduced the iso-fs module.Can contain either a mixture of div and
gap elements (typical for collections of samples), or
or a sequence of ab and gap elements (for
single texts). AP: this is a _clean_ modification. AP (090708):
reintroduced u, for spoken data. AP (090721):
reintroduced sequence of p.Used for collections of texts. Can contain a mixture of
ab and gap elements. Use the decls
attribute to refer to the appropriate place in the header
(typically an item element) that bears the metadata for
the given sample. The n attribute can be used to hold
the name of the number of the particular div, if
necessary. AP: this is a _clean_ modification. AP (090721):
reintroduced sequence of p.The sequence between two newlines (or between sequences of
newline+empty line, depending on the format of the input); these
blocks become identified and headings, paragraphs, etc., in the
structural annotation layer (ann_structure.xml). This element
can only contain plain text with optional gap or
hi elements (which are sometimes present in legacy
texts). AP: this is a _clean_ modification.Possible values are: "balanced" (for text in the 300-million-word balanced subcorpus), "unbalanced" (for other texts which may be distributed by NKJP), "restricted" (for texts available only for the internal NKJP purposes, not to be distributed), "one_million" (for texts in the 1-million-word manually annotated sample).restrictedWhich file (level) does the extent refer to? (Possible
values are text.xml, ann_segmentation.xml, etc.)text.xmlContains feature and feature-value libraries.The topic of spoken conversation. It might seem that
textDesc/domain could be used for this pupose, but, alas!,
inclusion of textDesc makes it necessary to use all of its 8
subelements (channel, constitution, derivation, etc.), while we
only need something like domain...The usual xml:lang attribute.