TEI for Better or Verse

1. TOWARDS A STANDARD FOR VERSE ENCODING?
The TEI P3 Guidelines (Sperberg-McQueen & Burnard 1994) give some
fairly open suggestions on how to use the tagset included for verse
encoding. With the use of the element in the TEI header many and
varied formalisms for encoding metrical and rhyme information can be
used. Since there is no (or almost no) standard formalisms expected
for this kind of data in the TEI no existing formalisms run the risk
of being excluded. On the other hand even the use of the
metDecl-information is not enough to easily allow for automatic
comparison and search for information on metrical and rhyme data in
data using different formalisms.

The project Henrik Ibsen's writings (HIW) is in the process of
encoding all works by Henrik Ibsen, including verse both in verse
drama, individual poems and verse parts found inside other genres.
HIW has developed an internal standard for verse encoding using the
TEI verse tag sets. This standard is also, we believe, useful for
others who encode, or would like to encode, (Germanic) verse. The
verse theory and methods are based on the metrical formalisms
developed by Jørgen Fafner (Fafner 1989).

As a result of discussions at two conferences organized by the
Nordic Network for Scholarly Editing (1999 and 2000 conferences)
work is now being done on a Nordic verse database using the same
TEI-implemented formalisms as the work done in HIW. Among the main
reasons for creating such a database are the possibility of
comparing verse and of looking for verse with a similar metric or
rhymic structure to the text one might be working with. We are
developing tools for search/retrieval and comparison of these data.

2. ON THE IBSEN MATERIAL AND VERSE - AND OUR WISH FOR INTERACTIVENESS
Henrik Ibsen is best known for his prose dramas. Early in his
career, however, he wrote several dramas either completely or partly
in verse. Ibsen has also written a large number of poems. The verse
material is large and difficult to overview, and there has been
modest research on the whole. An electronic, searchable verse base
with possibilities for comparative analysis would be very useful for
this material.

3. POSSIBILITIES FOR MULTI-LANGUAGE VERSE BASES?
The TEI Guidelines offers some advice and possibilities for markup
of verse form (div, lg, l etc.), but give little advice for a more
detailed notation system for verse analysis. This is understandable,
the TEI standards is mostly fairly open and so will not exclude
varying practices. Verse is also very much depending on linguistic
and historical features, so an encoding formalism useful for Danish
and Norwegian poetry from the 19th century will not be very useful
for Danish and Norwegian poetry from the 11th century not to mention
poetry written in other languages or language families and within
other poetic styles. But it may be possible to create a stricter
standard, or advice on how to use TEI for a restricted set of verse,
e.g. Scandinavian verse from the renaissance to modern times or
maybe even including all verse in Germanic languages from this
period. For searching and comparative analysis the encoding
formalisms used for verse texts should be as standardized as
possible.

At the conference we will demonstrate a TEI/XML based electronic
archive of verse by mainly danish and Norwegian authors. In addition
there will be samples of English and German poetry and some
non-Germanic texts.

4. FORMAL DESCRIPTIONS OF VERSE
The aim for our project's (HIW's) general markup is to store
information about the texts with all their structural divisions
based on the typographical information available in the source texts
rather than on our impressions of structural divisions based on our
analysis of text content. For the verse material we add a specific
verse markup based on an analysis of the verses. The verse encoding
is then divided in two main parts, one typographic and one analytic.
The first part includes recording all occurrences of verse (vs.
prose), divisions, linebreaks, indents, emphasizing etc. The second
part implicates descriptions of genre, meter, rhyme scheme, stress
pattern etc.

The information on verse structure in a text is encoded in general
formulae located at the highest possible level within the textual
hierarchy (as recommended by TEI). Lower-level elements will inherit
the properties defined above, and any deviation from the general
verse system will be specified at the level where it occurs.

Our metrical formulae are based on the work done by Jørgen Fafner,
and his "Danish Metrics" (Dansk Verskunst), 1989-. Fafner has based
his notation on the arrangement of stressed and unstressed
syllables, which is the basic method for creating metrical
structures within the Germanic languages.

5. EXAMPLE OF VERSE NOTATION
One of Ibsen's earliest poems, "Resignation", may serve as an
example of how Fafner's notation is used:
RESIGNATION (1847)

Er de Glimt fra Sjælens Dunkle,
Der igjennem Mulmet brød,
Og som Lynblink monne funkle
Kun til evig Glemsel født? -

Var forgjæves al min Higen,
Var min Drøm kuns et Fantom,
Er mig nægtet Sjælens Stigen,
Var min Digten kold og tom! -

Tier da I Undertoner! -
Kan jeg eder ei forstaa, -
Lad mig iblandt Millioner
Leve glemt og glemt forgaa! - - -
The main verse information is stored in this formula:
"bi (8 7)2 (Ab)2"

The abbreviation 'bi' is short for 'bisyllabic' which tells us that
each measure consists of two syllables, one stressed and one
unstressed. The numbers (8 7) show the number of syllables in each
verse, and the figure two outside the parenthesis means that the
number of verses should be multiplied by two. Each stanza will then
include four verse lines. The letters (Ab) visualize the rhyme
pattern in the verses. Female and male rhymes are specified with
respectively upper case and lower case letters.

Our challenge is to include Fafner's notation in our electronic
texts according to the TEI Guidelines. The TEI Guidelines recommends
to markup larger units, such as poems in 'div' elements, verse lines
with the 'l' element, and groups of verse lines with the 'lg'
element. Smaller units than verse lines may be marked up by the
'seg' element. All these elements, except 'div', may contain the
attributes 'type', 'met', 'rhyme' and 'real' to specify respectively
genre/form/meter, metrical convention, rhymes and metrical
deviations.

We realized early that chapter 9 in the TEI Guidelines is a solid
starting point, but that the encoding has to be developed further
for us to be able to describe the often complicated verse structures
in actual texts. The poem "Resignation" has a simple verse
structure, which causes no problems when filling in the 'met' and
'rhyme' attributes. Based on Fafner's notation system we can define
the metrical convention for this poem as 'met="bi (8 7)2"'. The
final rhymes are tagged with the attribute 'rhyme="(Ab)2"'. So far
we have not encoded alliteration and internal rhyme, but this is not
a final decision.

Encoding problems arise with verse texts with various degrees of
variations and deviations. The TEI Guidelines offers no attribute
for showing deviations from the rhyme scheme. David Chisholm and
David Robey suggest (Chisholm & Robey 1995) creation of a new
attribute, 'realRhyme'. We have applied this solution, and have also
replaced the 'real' attribute with the new attributes 'realMet' and
'realRhyme'. There are also texts with more systematic variations,
as when female and male rhymes alter within a fixed rhyme scheme.
The number of syllables may also be varying, both within the
measures and within the verse lines. And there is not always a
regular pattern for anacrusis. In our texts we wish to include both
the general verse pattern and the variations and deviations from
this. We have discussed several alternatives for solving the
problems described above, and we will present our conclusions, along
with a more systematic overview of verse encoding at our project, at
the conference.

Bibliography:
Chisholm, David & David Robey 1995. "Encoding verse texts".
Computers and the Humanities, 29, 99-111.
Fafner, Jørgen 1989-2000. Dansk verskunst. Copenhagen. 3 volumes.
1. Digt og form: Klassisk og moderne verslære. 1989.
2. Dansk vershistorie 1: Fra kunstpoesi til lyrisk frigørelse. 1994.
3. Dansk vershistorie 2: Fra romantik til modernisme. 2000.
Sperberg-McQueen C.M. & Lou Burnard 16/5 1994. Guidelines for
electronic text encoding and interchange (TEI P3 Guidelines).