Encoding journals and other serial publications

A number of decisions need to be made when using the TEI to encode digitized journals or other serial publications. Even once those decisions are made, there are still a few options for how you might encode these. This document attempts to summarize those options.

It might be tempting to use the <teiCorpus> structure for journal issues or volumes in order to allow a separate <teiHeader> not only for each article but also for each issue, volume, and the journal as a whole. This element is meant for language corpora or collections of encoded texts, with the unstated assumption that the collection is heterogeneous. So it's not really appropriate.

But here are some other options:

Unit of text for each TEI document	Advantages	Where to put metadata for the units of text that do not correspond to the TEI document
Option A: One article per TEI document	General-purpose digital library systems will easily be able to include individual articles in search results, though possibly without the context of articles in the same section or of the containing issue, volume, and serial title.	You can have a <bibl>, <biblStruct>, or <biblFull> in the <sourceDesc> for the article. It will be akin to an analytic catalog record which includes not only metadata about the article but also references the journal title, volume, and issue.
Option B: One issue per TEI document	An issue may contain front and back matter worth preserving that doesn't fit into the model of a journal article very well. It may also contain an introduction that refers to other articles. Preserving this context is simpler if each issue is its own TEI document: front matter would go in <front>, back matter in <back>, and each article including the introduction could be encoded as a <text> within a <group> or as a div within a single <text>.	Metadata specific to the article would need to be found in the <text> or div for that article, but including metadata beyond a transcription of what's found in the source document somewhere within that <text> or div can introduce complications in the TEI encoding. You either would need to have more than one element for each component (one that's a transcription and one that's the normalized form), or you would need to group the normalized forms somewhere within the <text> or div in an element like <bibl> not really designed for this purpose.
Option C: One (bound) volume per TEI document	If your project digitizes bound journals like monographs, you may have a set of page images corresponding to the whole volume and want to keep your TEI document at this level as well.	Article-specific metadata would be treated as with option (B). In addition, there are two ways to handle metadata for higher-level units: You can have metadata for the volume and issue in the appropriate place in the <text>, but including metadata beyond a transcription of what's found in the source document can introduce complications in the TEI encoding. You either would need to have more than one element for each component (one that's a transcription and one that's the normalized form), or you would need to group the normalized forms somewhere within the <text> in an element like <bibl> not really designed for this purpose. You can have a single <bibl>, <biblStruct>, or <biblFull> for each volume or issue within the <sourceDesc>. The <text> or div for each article would include a @decls pointing to the appropriate <bibl>, <biblStruct>, or <biblFull>.