rhyme schemes

[David J Birnbaum]

Dear SEELANGers,

I am currently working with a computer science colleague on an automated
poetry analysis project, which takes an electronic file of Russian verse
and generates a report on the rhyme scheme. Our report needs to record
four things:

1) which lines rhyme;
2) whether rhymes are masculine, feminine, dactylic, etc.;
3) whether rhymes are perfect or imperfect;
4) in the case of imperfect rhymes, the nature of the imperfection(s).

My current inclination is to generate a report of the following type:

A 2 01000100       blah blah blah

The first column ("A") will list A, B, C, etc., in conformity with the
normal way for expressing rhyme schemes (e.g., ABAB, ABBA, etc.). If the
rhyme is perfect, the letter will be uppercase; if it is imperfect, the
letter will be lowercase. The second column ("2") indicates the depth of
the rhyme; 1 = masculine, 2 = feminine, etc. The third column is an
eight-digit binary number (which we can make longer, if necessary), where
each digit represents the presence ("1") or absence ("0") of a particular
type of imperfection (imperfect correspondence of post-tonic vowels,
post-tonic consonants, etc.). All fields will be "0" in case of perfect
rhyme. Then comes the line of poetry ("blah blah blah").

I have seen different conventions for indicating different types of rhyme.
If I remember correctly (I don't have the book here to check), Nabokov, in
his "Notes on Prosody," uses uppercase consonants for masculine, uppercase
vowels for feminine, and lowercase for long (dactylic or longer) rhymes.
But he doesn't mess around with imperfections, which are important to us.

Can any versification experts out there offer suggestions about the most
useful reporting formula? I would like a system where a short report (such
as one omitting the binary field) would provide most of what a student of
poetry requires, but a longer report (including the binary field) will be
available for finer analysis. The information scientists among you will
have noticed that the distinction between uppercase and lowercase letters
is inferrable from the binary field, and is therefore redundant, but I'd
like to keep it for purposes of generating this type of short report
easily (by simply suppressing the binary field). The reporting formula
should conform, to the greatest extent possible, with conventions that
will be familiar to versification specialists.

I would also be grateful for help in identifying types of imperfect rhyme.
I have a list of different types of imperfections, which I extracted from
chapter 6 of Unbegaun's handbook of Russian versification and supplemented
by reading Esenin, but if anyone who is familiar with Russian rhyme wants
to send me a list of imperfections, I'll be happy to include any I may
have missed into the system.

I currently make two assumptions: rhyme must involve identity of the
stressed vowel sound and rhyme must be isosyllabic. The latter assumption
is false for Majakovskij (who rhymes <gorode> with <morde>) and others,
and we'll expand the system to deal with heterosyllabic rhyme as soon as
we can figure out how to constrain the possible types algorithmically (if
anyone has an electronic file of Majakovskij's poetry to share--in any
encoding--this would help!). (We know about supporting consonants in open
masculine rhymes; we treat the absence of a supporting consonant, which
we've encountered in Esenin, as imperfect rhyme, rather than absolute
non-rhyme.)

For those who are interested, the source code (in C), an input file
(several hundred lines of Esenin in George Fowler's encoding), and
documentation will go up on my web page in a couple of weeks, along with a
report on development. I'll try to get binaries compiled for OS2 and DOS
as well, but no promises there. Copyright by the authors, but free for
noncommercial use. I'll post an announcement here when the time comes.

Cheers,

David
==================================================
Professor David J. Birnbaum      djbpitt+ at pitt.edu
The Royal York Apartments, #802  http://www.pitt.edu/~djbpitt/
3955 Bigelow Boulevard           voice: 1-412-624-5712
Pittsburgh, PA  15213  USA       fax:   1-412-624-9714