From 74774.157@compuserve.comThu Mar 21 16:12:29 1996
Date: 20 Mar 96 22:52:17 EST
From: Jeffrey Henning <74774.157@compuserve.com>
To: BlindCopyReceiver:  ;
Subject: MODLANG 8 (2 of 2)

oVolume I, Issue 8 -- January-February, 1996 (Continued)


EMULATING TOLKIEN

When learning a craft, whether writing, sculpting or creating model 
languages, it always helps to begin by purposefully copying a master's 
style.  This enables you to begin creating and gives you time to 
experiment prior to developing your own style.  For instance, my first 
poetry slavishly followed e.e. cummings.  My first attempt at a 
children's book faithfully echoed the voice of Dr. Seuss.  
Interestingly, my first attempt at a model language did not emulate 
Tolkien, but copied Clyde Heaton's Orcish, a language he published in an 
article called "Even Orcish is logical" (_Dragon_ magazine, July 1983), 
which started me on this hobby.

One of the language families I have tinkered with the longest is meant 
to be spoken by elves, like Tolkien's Quenya and Sindarin;  my Alvish 
and Old Alvish languages date back some 13 years now.  My first attempt 
at Alvish was patterned closely on Anglo-Saxon, and the language existed 
as little more than a source language for Karkrak, then the principal 
language I was working on.  About five years ago, I revamped Alvish to 
resemble ancient Greek, which I first thought gave the language a noble 
sound.  When that did not meet my fancy, I revamped the language again 
and developed a 3000-word vocabulary for it.  Some sample vocabulary, 
from Alvish III:
 
   age, n.            thufpaxef, neut.  (thoofPAHKSef)
   age, v.            thufpaxere, neut.  (thoofpahksEre)
   agree, v.          cupfuscrifere, neut.  (koopfooskriFEre)
   agreement, n.      cupfuscrifep, neut.  (koopfoosKRIfep)
   agriculture, n.    mudolrelep, neut.  (moodolRElep)
   ahem, interj.      ixax  (IKSahks)
   ahoy, interj.      esathnebis  (eSAHTHnebis)
   aid, v.            ethere, neut.  (Ethere)
   air, n.            tusef, neut.  (TOOSef)
   alarm, v.          esadadere, neut.  (esahDAHdere)
   alas, interj.      frithfrith  (FRITHfrith)
   alienate, v.       apadere, neut.  (ahpahDEre)
   alignment, n.      fadap, f.  (FAHdahp)
   all, adj.          quafeth, neut.  (QUAHfeth)
   all, adv.          quaferemi, neut.  (quahFEremi)
   all, n.            quafep, neut.  (QUAHfep)
   alleluia, interj.  ipipneb  (Ipipneb)

I have no idea what I was aiming for, but the result is undeniably ugly.  
Ipipneb?  Frithfrith?

Recently, I have decided to try and pattern Alvish more closely on 
Quenya and Sindarin, inspired in part by a recent discussion on the 
CONLANG mailing list, where David Bell discussed his language Amman-lar, 
originally a Tolkien clone, before Bell found his own voice (you can 
subscribe to CONLANG by sending an e-mail with the subject text 
SUBSCRIBE CONLANG <YOUR NAME> to majordomo@diku.dk).

Before creating Alvish IV, I closely studied the Quenya and Sindarin 
words published in _The Silmarillion_.  I decided that what sounded 
pleasing to me was the emphasis on sounds produced towards the front of 
the mouth (e.g., /p/, /b/, /f/, /v/, /i/, /e/, /a/).  Additionally, 
syllables in Sindarin and Quenya typically followed the style CV or CVN, 
where N was a nasal (/m/ or /n/), lateral (/l/) or approximant (/r/).  
Based on this insight, I then produced the following definition of the 
phonotactics (sound and syllable structure) for Old Alvish, the ancestor 
of Alvish:

   (Con1) Vwl (Con2)
   
   where 

   Con1 = p, b, f, v, t, d, c, g, y, w, ch, gh
   Vwl = i, a, o, u, uu
   Con2 = m, l, n, r, s

The /c/ is pronounced as in _cat_, /ch/ as in _loch_, /gh/ is a voiced 
/ch/ (as /g/ is a voiced /c/).  Each of the above lists of phonemes is 
roughly arranged with those sounds pronounced closest to the front of 
the mouth listed first.  Sounds are listed in declining order of 
occurrence in actual Old Alvish words.

A further restriction to possible combinations of phonemes is that 
whenever there are two adjacent vowels (e.g., the /i/ and /o/ are in 
separate syllables in _dios_, /di-os/) they can only be one of the 
following:  i-o, i-uu, i-a, u-o, u-i, u-a, o-i.  So, for example, _diis_ 
is not a valid Old Alvish word, since /i-i/ is not a valid combination.  
(The dash - is used to indicate syllable breaks.)

Some sample Old Alvish words and phrases:
   anim
   basuus
   buci
   chi gicuu
   dafon
   ghis
   gibas
   ogus dian
   toman
   vafus

The sound system for Old Alvish is less flexible than that of Tolkien's 
languages.  Words like _Aglarond_, _Amarth_, _Bragollach_ and _Minas 
Tirith_ could not be formed in Old Alvish, due its different syllable 
structure, but would be probably be borrowed in forms like _Agelaronde_, 
_Amarte_, _Beragolcha_ and _Minas Tirti_.

But Old Alvish's sole purpose in my design is to provide a source for 
Alvish.  I designed Old Alvish to have an elegant structure, which has 
become more complicated (and therefore more flexible) in Alvish.

I decided that Alvish developed from Old Alvish according to the 
following steps.  First, Middle Alvish was distinguished from Old Alvish 
by the following sound shifts, designed to give the language even more 
front sounds:

   /c/  > /th/
   /g/  > /h/
   /ch/ > /c/
   /gh/ > /g/

   /a/  > /e/
   /uu/ > /o/
   /o/  > /a/

This had the net effect of giving Middle Alvish three sounds that it did 
not have before (/th/, /h/ and /e/) while eliminating three sounds that 
it had had (/ch/, /gh/ and /uu/).

Some examples:

   MA buthi < OA buci
   MA ci hitho < OA chi gicuu
   MA defan < OA dafon
   MA gis < OA ghis

As this example should illustrate, you can quickly generate your own 
languages, based on a source language, simply by preparing tables of 
sound correspondences such as the one above.  (Refer to the "Sounds" 
section of MODLANG 2 for more information on this.)  A phoneme can 
gradually become pronounced as any "neighboring" sound, where a 
"neighborhood" consists of similar physical positions of articulation.  
By this measure, the shift of /c/ to /th/ is unlikely, as the physical 
positions of the two sounds are far apart, but this can be explained 
away by positing an intermediate step;  e.g., the sound /c/ came to be 
pronounced as /t/ (as in OA _buci_ becoming Early Middle Alvish _buti_), 
before the /t/ phonemes so produced came to be pronounced as /th/ (MA 
_buthi_).

The next series of sound shifts distinguishes Middle Alvish from Alvish 
and is more complex.  In this series, there are no straightforward one-
to-one correspondences, where one occurrence of a phoneme always becomes 
another phoneme.  In the transition to Alvish, sounds changed only 
because of their environment (the other sounds they are pronounced 
near).  While the notation used to describe these can grow quite 
complex, inventing sound changes like the following is not difficult.  
Basically, I spent a lot of time trying to make sure that Alvish words 
fit my preconceptions of what words I found aesthetically pleasing, and 
I then formulated rules to give me a way to get from the strict 
phonotactics of Old Alvish to something looser.

The main changes from Alvish to Old Alvish are in the phonotactics.  An 
Alvish word can begin or end with any consonant, but the consonants in 
the middle must follow similar patterns to Old Alvish.

   FIRST SYLLABLE
      (ConWI) Vwl (ConSF)
   INTERNAL SYLLABLES
      (ConSI) Vwl (ConSF)
   TERMINAL SYLLABLE
      (ConSI) Vwl (ConWF)

   ONE-SYLLABLE WORD
      (ConWI) Vwl (ConWF)

   where 
      ConWI = p, b, f, v, t, d, c, g, y, w, th, h, m, l, n, r, s, sp, 
st, sc
      ConSI = p, b, f, v, t, d, c, g, y, w, th, h, s, sp, st, sc
      Vwl = i, e, a, u, o
      ConSF = m, l, n, r
      ConWF = p, b, f, v, t, d, c, g, y, w, th, h, m, l, n, r, s

Clearly, this sort of mapping out of all sound combinations can grow 
much more complex than you want or need for a model language, but the 
complexity of Alvish phonotactics pales in comparison to the 
phonotactics of English, which would take a small book to describe in 
detail (e.g., the only time an English word can begin with three 
consonant sounds the first consonant has to be /s/, as in _spring_, not 
something like _zbring_).

The rules for deriving Alvish words from Middle Alvish are:

   /ti/ > /thi/
   /a-i/ > /i-a/ (eliminating the only vowel pair to begin with /a/; 
e.g., /thi-al-fu/ < MA /tha-il-fu/)
   /s-/ > /-s/ (removing /s/ from being a possible final consonant for 
internal syllables;  e.g., /i-san/ < MA /is-an/)
   /-sX/ > /-X/ where X<>/a/,/e/,/i/,/o/,/u/,/p/,/t/,/c/  (any /s/ that 
migrated before an X is omitted;  e.g., /pa-fu/ < MA /pas-fu/)
   /-DVC#/ > /D#/ where D=/t/,/d/; V is a vowel, C is consonant and # 
indicates the end of a word (this rule indicates that final syllables 
that begin with /t/ or /d/ move the dental phoneme to the end of the 
previous syllable and truncate the remaining syllable;  e.g., /alt/ < MA 
/al-ten/)
   /mD#/ > /nD#/  (any final /md/ or /mt/ formed by the previous rule 
replaces /m/ with /n/)
   /#V1T-V2/ > /#TV2/ where T=/m/,/n/,/l/,/r/,/s/ (any word that begins 
with a vowel and is followed by a syllable-terminating consonant and a 
following vowel -- starting a new syllable -- drops the initial vowel 
and moves the consonant to the next syllable;  e.g., /nim/ < MA /en-
im/).

One thing I've ignored when doing sounds shifts is whether a phoneme was 
in a stressed or unstressed syllable.  Contrast English /ob-JECT/ ("I 
object, your honor") to /OB-ject/ ("the object of the game is this").  
In English, the vowel /o/ has become a schwa in the unstressed syllable 
in /ob-JECT/ but not in the stressed syllable in /OB-ject/.  Rather than 
deal with the issue of stress (which in the Alvish languages always 
falls on the penultimate -- next-to-last -- syllable), I decided that 
elves are more fastidious in their pronunciation and blur sounds less in 
unstressed syllables than mere mortals do...  It's a cop out, but this 
is supposed to be fun, right?

I won't formally describe the rules for forming compound words in 
Alvish.  But forming compounds does have a number of twists, mainly 
designed to make sure that the resulting word matches the phonotactics 
of Alvish defined above.  The rules:
*   For the first of the two words compounded, anytime a vowel 
immediately follows another vowel (e.g., /e/ in /i-e/), that vowel is 
dropped.  
*   When the first word ends in a vowel and the second word begins with 
a vowel, those vowels change into an acceptable vowel pair (if they're 
not already); e.g., i-a, i-o, i-e, u-a, u-i, u-e.  
*   When the first word ends in a consonant, that consonant can be only 
/m/, /n/, /l/ or /r/, otherwise it moves to the front of the second word 
(if the second word begins with a vowel) or it is deleted.  
*   This process is inverted when the second word *begins* with a 
consonant;  since that consonant cannot be /m/, /n/, /l/ or /r/, the 
consonant moves to the end of the first word, unless the first word ends 
in a consonant, in which case it is dropped altogether.  Whew!

Based on all these rules, I derived some sample words for Alvish.  These 
words have no assigned meanings, since I have just been experimenting to 
make sure that I like the resulting sounds of the words generated.

   a-hu-di-en [< a-hus di-en < OA o-gus + di-an.]
   a-pun [< OA o-pun.]
   alt [< MA al-ten < OA ol-tan.]
   as [< OA os.]
   ban-u [< OA bon-u.]
   be [< OA ba.]
   be-sos [< MA bes-os < OA bas-uus.]
   bel [< OA bal.]
   bet [< MA be-tu < OA ba-tu.]
   bi-al [< MA ba-il < OA bo-il.]
   bi-hi-bes [< OA bi + gi-bas.]
   bi-om [< OA bi-uum.]
   bo-ses [< MA bos-es < OA buus-as.]
   bol-i-e-ba [< bol-e e-ba < OA buul-a + a-bo.]
   bot [< MA bo-thar < OA buu-cor.]
   bul-a-sim [< bu-el a-sim < MA bu-el as-im < OA bu-al + os-im.]
   but [< MA bu-thi < OA bu-ci.]
   ci-hit [< MA ci hi-tho < OA chi + gi-cuu.]
   de-fan [< OA da-fon.]
   di-os [< OA di-uus.]
   di-si [< di-as i < OA di-os + i.]
   dir-a-bi-en [< di-ar-a bi-en < OA di-or-o + bi-an.]
   du-an [< OA du-on.]
   du-ha [< OA du-go.]
   e [< OA a.]
   e-bun-bid [< MA e-bun bi-da < OA a-bun + bi-do.]
   e-pa-tam-en [< OA a-po + tom-an.]
   em-ho [< OA am-guu.]
   fa-po [< OA fo-puu.]
   far [< OA for.]
   fe-pu [< OA fa-pu.]
   fe-vin-ve-fus [< OA fa-vin + va-fus.]
   fel-im-fon-et [< fel-i-am fon-et < MA fel-a-im fon-e-tul < OA fal-o-
im +  fuun-a-tul.]
   fes [< OA fas.]
   fir-o [< OA fir-uu.]
   fu-en [< OA fu-an.]
   gi-su [< gis u < OA ghis + u.]
   hit [< MA hi-tan < OA gi-ton.]
   i-ba [< OA i-bo.]
   i-fen-pur [< OA i-fan-pur.]
   i-san [< MA is-an < OA is-on.]
   ir-u-se [< MA ir us-e < OA ir + us-a.]
   ni [< MA on-i < OA uun-i.]
   nim [< MA en-im < OA an-im.]
   nur [< MA en-ur < OA an-ur.]
   o-fam [< OA uu-fom.]
   pa-fu [< MA pas-fu < OA pos-fu.]
   pa-fu [< OA po-fu.] (arrived at the same form as the previous word 
but by a different route)
   pe-yil [< OA pa-yil.]
   pi-em-do [< OA pi-am-duu.]
   pi-fem [< OA pi-fam.]
   pi-sun-bi-as [< MA pis-un ba-is < OA pis-un + bo-is.]
   pir-di-fus [< pi-ar di-fus < OA pi-or + di-fus.]
   po-fu [< OA puu-fu.]
   po-vi [< OA puu-vi.]
   pod [< MA po-des < OA puu-das.]
   pom-i [< OA puum-i.]
   pu-em [< OA pu-am.]
   pu-il [< OA.]
   pul-on [< OA pul-uun.]
   ran [< MA er-an < OA ar-on.]
   te [< OA ta.]
   ter-in [< OA tar-in.]
   the-wom-o [< OA ca-wuum-uu.]
   thi [< OA ci.]
   thi-al-fu [< MA tha-il-fu < OA co-il-fu.]
   thi-as [< OA ci-os.]
   thin-o [< thi-on no < MA thi-on an-o < OA ci-uun + on-uu.]
   tho-pa [< OA cuu-po.]
   thot [< MA tho-tol < OA cuu-tuul.]
   ti-po-dom-pi [< OA ti-puu + duum-pi.]
   ti-vol [< OA ti-vuul.]
   tim-vul-po-wim [< OA tim-vul + puu-wim.]
   tin [< OA.]
   to-por [< OA tuu-puur.]
   tu-es [< OA tu-as.]
   u-fi [< OA.]
   u-va [< OA u-vo.]
   um [< OA.]
   urt [< MA ur-thi < OA ur-ci.]
   ut [< MA u-tho < OA u-cuu.]
   va [< OA vo.]
   val-i [< OA vol-i.]
   vam [< OA vom.]
   vem [< OA vam.]
   vi-al [< MA va-il < OA vo-il.]
   vi-fom-fe-hu [< OA vi-fuum + fa-gu.]
   vim-u [< OA.]
   vo-fi [< OA vuu-fi.]
   vod [< MA vo-di < OA vuu-di.]
   wil-pe [< OA wil-pa.]
   ya-wir-os [< OA yo-wir-uus.]
   yar-o [< OA yor-uu.]
   ye-bon [< OA ya-buun.]
   yed [< MA ye-dum < OA ya-dum.]
   yi-si-pu [< yis i-pu < OA yis + i-pu.]
   yo [< OA yuu.]
   yu-es [< OA yu-as.]
   yu-pi [< OA.]

Some sample sentences (really just random sequences of words):

   Ut ofam tivol pomi fes te pafu firo vial fevinvefus.  
   Bihibes fepu yupi be thopa biom ahudien vali yed puem.  
   Vifomfehu wilpe iba apun iruse piemdo thino duha bulasim fapo.  
   Ut tipodompi dirabien boses bial e terin yebon ufi ni.  
   Vam va dios thewomo thi defan urt bot ifenpur uva.  
   Fuen pofu vem as nur bet pirdifus ebunbid tin bel.  
   Pifem hit isan epatamen tues nim duan bolieba vofi pisunbias.  
   Disi vimu banu cihit pod thialfu pulon yawiros besos yaro.  
   Yo pafu ran peyil timvulpowim puil alt thias topor um.  
   Yisipu be felimfonet thot emho vod povi yues far gisu.  

So there you have the sounds of Alvish.



***



You have just finished reading _Model Languages_, a regular on-line 
newsletter published 12 times a year and provided free to all interested 
parties.  Feel free to post this newsletter on BBSes or online services 
and feel free to e-mail it to others, so long as you forward it in its 
entirety.  To become a regular subscriber, send a message with the text 
"SUBSCRIBE MODLANG \8 [your name] (your http address)" in the subject 
header to 74774.157@compuserve.com (include your http address only if 
you would like your home page to be linked to the _Model Languages_ 
membership directory home page).  To cancel a subscription, send a 
message with the words "UNSUBSCRIBE MODLANG" in the header.  I look 
forward to all comments, including corrections, and am always interested 
in possible articles for inclusion in future issues.

To review back issues, visit the _Model Languages_ home page at 
http://ourworld.compuserve.com/homepages/jeffrey_henning.

Contents copyright 1996 Jeffrey Henning.  All rights reserv