Conlang optimized for a specific text

As I was driving home from my brother's house I reflected, not for the
first time, how unwieldily long-winded the Hail Mary is in gjax-zym-byn.
Somehow I then thought of a conlang design method oriented around
a specific text -- the initial core design is made to express the meaning
of said text in the most concise way (or pretty concisely while preserving
some degree of euphony, or whatever) -- then the design decisions made
with respect to that text are extrapolated in some reasonably consistent
way in fleshing out the language for general-purpose expressivity.
For instance, most or all of the concepts required for the text would
get a monosyllabic word (or a one-phoneme affix), while concepts
that do not occur in the text, even if they would be fairly common in
general use, would have to make do with the word-shapes left over,
many of them disyllabic or longer.  For the sake of concision some
words used several different ways in the text might get irregular
suppletive inflection for case, number or whatever while words not
used in said text would mostly have regular agglutinative inflection.

I've started working this out in detail with the prayers of the Rosary,
starting with the Hail Mary (since it's used most often) and then
the Our Father; the other prayers will follow, then misc translation
exercises without the conciseness goal to flesh out the
rest of the language.  The Hail Mary is  27 syllables in this
mini-conlang, compared to 67 syllables in Latin by my count,
or 50 in English.  Part of it:

mar ja li-n tu m' be, po be l' da-m s' po jesu ro f' ti d' be.
/mar\.ja.lin.tum.be.po.bEl.dam.spo.jE.su.r\of.tid.be/

Mary.VOC hail grace-FUL lord with you be_blessed you among woman-PL
and be_blessed Jesus fruit out_of womb of_inalienable you

The prepositions and conjunctions are all consonant+schwa,
and the schwa is elided whenever a permitted consonant cluster
would ensue; of course the word forms in this text are chosen
so that schwa is almost always elided.  Since vocative names
occur more often than nominative names here,
the vocative of proper names is formed by truncating the
final vowel of the nominative (or absolutive?) form.  However,
I might get rid of the nom/voc. case distinction in the final revision
since I wound up not finding a need for other syntactic cases
in the rest of the text; strict VSO word order seemed to make
it all clear enough.

Between the Hail Mary, Our Father and Glory Be,
I've coined 62 morphemes (58 words and four suffixes), using up
about two-thirds of the space of 95 possible CV words with my
phonology.  (All consonant clusters result from affixes
and clitics in running text; roots are CV(CV)*. )  New root
morphemes needed for the the Apostles' Creed will
get disyllabic CVCV forms, I expect.  Then I'll start in on some more
translation without the obsessive conciseness goal,
and some original writing including additional example sentences
for morphemes that occur only once in the Rosary corpus.

Have any of y'all ever designed a conlang around a target
text like this?

--
Jim Henry
http://www.pobox.com/~jimhenry

Reply