Hallo conlangers!

On Sunday 27 January 2013 17:39:48 Melroch wrote:

> I'm afraid this didn't go to the list the first time I sent it.
> 
> ---------- Vidarebefordrat meddelande ----------
> Från: *Melroch*
> Datum: söndagen den 27:e januari 2013
> Ämne: [CONLANG] Using pre-existing conlang Roots to create a new language
> 
> 
> The actual unconditioned chain shift in Old High German is pretty cool:
> 
> d > t > T > d
> 
> except that the outcome of single *t is actually [s_m] after vowels and
> [ts_m] elsewhere including when geminated.

And */T/ > /d/ predates the rest, as a /d/ resulting from this
is devoiced in Upper German.  Also, it has happened in Low German,
too.
 
> I'd try to come up with some earlier perhaps conditioned shift which feeds
> such a shift.
> 
> I've done my share of bogolanging of the "apply the sound changes of
> natlang X to natlang Y" variety and found that it only brings you so far.

Indeed.  I have made *one* language of that kind, Roman Germanech,
which has some strong points (I particularly like how the interplay
of Early Romance and German sound changes causes VL */E/ and */O/
to leap-frog */e/ and */o/ and become /i/ and /u/, respectively);
but I think I need not try again.  The general problem with this
is that natlang X and natlang Y almost never have the same
phonology and thus the sound changes of natlang X make little sense
when applied to natlang Y.  You must bend them creatively to make
them make sense!

>        I
> have also done my share of "apply the sound changes of natlang X to
> a-priori conlang Y" and above all the "what sound changes do you need/can
> you apply to con-/natlang X to make it resemble natlang Y" variety. The
> thing is that I almost always end up spicing up the stew by introducing
> something from natlang Z, related to neither X nor Y.

I attempted some such conlangs, applying the sound changes of this
or that European language to Proto-Hesperic, but it failed to work
out to my satisfaction, chiefly because Proto-Hesperic has a quite
different phonology than PIE.

As a consequence, I now use my own sound change lists in my work
with the Hesperic languages, even if they are in some instances
*influenced* by natlang sound changes.  For example, Proto-
Alpianic has undergone a consonant shift similar to the extreme
form of the High German shift that happened in Swiss German, but
in many other points its sound changes are very different from
those of Swiss German!

>        Take for example
> Rhodrese, which was in origin a renewal of a project from my teens taking
> bits from Italian and French and build a Romance language most pleasing to
> me.

Oh, weren't things easy when we weren't yet burdened with the kind
of linguistic knowledge we have now? ;)  But on the other hand, we
now have much more fun with the diachronic method!

>       Of course I know more now so I'd be more careful of making it a
> plausible descendant of Vulgar Latin, and I would take care not to get too
> close to Occitan (for all my admiration for that language) Lombard or
> Romantsch. So I suddenly find myself introducing Germanic-style i-umlaut
> and then bending the pattern of the development of nominal plural endings
> in Italian to get a starred system where all three declensions have plurals
> in _*-i_ which feed i-umlaut.

I think that can be done without administrating violence to Vulgar
Latin, so why not?

>       I don't know if such things make such
> conlanging more 'advanced' but it does make it more fun, and more "a priori
> in a posteriori" which I guess is a Good Thing. After all the whole point
> of Rhodrese is to create a romlang which satisfies my possibly warped
> aesthetic sensibilities -- a yummy lámatyáve as Tolkien would probably not
> have said.

At least, it does some things differently than other Romance
languages, and that is a Good Thing.
 
> With Sohlob I've tried to work backwards to come up with an earlier stage
> of an (a priori) conlang which would allow me to craft realistic sound
> changes to come up with sister languages with certain characteristics,
> mostly inspired by certain natlangs. This has proven harder than I
> expected, mainly because the characteristics of the daughter languages are
> largely already predetermined and it becomes a requirement on the mother
> language to be able to realistically change into all those daughter
> languages.

I know what you are talking about.  Parts of Proto-Hesperic are
made up backwardly from Old Albic; others are made forwardly from
Proto-Europic, which was made up backwardly from PIE.  It is not
always easy to get these strands together; of course, I could
change Old Albic to fit, but I try to find solutions which get by
with as little change to Old Albic as possible (in practice, this
has shown to work quite well; also, that way I find occasion to
add flesh to the sound change list of Old Albic, which I sometimes
feel to be too short for the time depth of the Hesperic family, let
alone Europic, which takes one even farther back than Hesperic).
And of course, I am not in any position to tweak PIE to fit ;)

>       E.g. the eastern Sohlob dialects which were originally were
> designed to phonologically resemble Turkic and modern Persian lack any form
> of quantity distinction in both vowels and consonants, and so did their
> mother language. Now I wanted the western Sohlob languages to resemble
> Germanic at least a little and so to have both long and short vowels. I
> could not very well just decide that the mother language had long vowels
> without upsetting the patterns of syncope and apocope which obtained
> between the mother language and the eastern languages. Instead I
> 'reconstructed' a grandmother language which had both long and short
> *consonants* between vowels and then the following sound changes in the
> 
> western branch:
> | Vd  > V:d > V:D
> | Vt  > Vd > VD
> | Vdd > V:dd > V:t
> | Vtt > Vtt > Vt
> | Vxt > Vtt

An interesting way of achieving this.
 
> originally consonant length was supposed to disappear without trace in the
> eastern branch but there was a small phonotactic weirdness in that branch:
> it didn't have any intervocalic voiceless stops since in the older version
> without quantity I had had
> 
> | VdV > VDV > VzV
> | VtV > VdV
> 
> to which I now added
> 
> | VddV > VdV
> | VttV > VtV
> 
> I decided that *dd had been pretty rare, but that the western branch had
> *zd > *dd before the quantity shift so as to make V:t less rare in the
> daughter languages. However I'm still not quite satisfied, as the relative
> frequency of V:D vs. V:t in the western branch and Vd vs. Vt in the eastern
> branch doesn't feel quite realistic. The probably best solution is to have
> /d/, /t/, /tt/ but no /dd/** in the mother language

Makes sense.  Perhaps in Pre-Proto-Sohlob, there were three degrees
of consonant length and no voicing, and the Proto-Sohlob system
emerged by a chain shift *ttt > *tt > *t > *d.  Kind of like the
consonant gradation systems in Balto-Finnic languages.

>        and then have
> west-Germanic-like Cj > CC feeding the quantity shift in the western
> branch. That would be easily introduced since +y clusters and
> palatalization are prominent in the eastern branch.

Yep.

>        This however imtroduces
> another problem in that palatalization of *ry was __the__ source of /l/ in
> both branches. There are two possible solutions: (1) unconditioned *r > l
> and *z > r in the western branch, which would make /rt/ strangely rare or
> (2) a late *l > *r and then even later *ry > l in the eastern branch, which
> at least is naturalistic since it actually happened in north Iranian (Latin
> _Alani_ borrowed from a reflex of proto-Iranian *āryāna.

This indeed makes perfect sense!

> Anyway thus it is a constant tug-of-war between tweaks to mother, daughter
> and sister languages. It _is_ fun but it rather shatters any hopes of ever
> rounding off the family tree. Which I guess is good. I don't think I want
> to get finished with this, but I feel that things morph too fast for me to
> keep my vocabulary db updated...

Certainly, such tinkering with sound changes is *a lot of fun*.
But indeed, it involves the risk of getting lost in a mess of
alternative solutions which one finds it hard to choose from.
It *is* useful to set some constraints, such as to treat some
of the conlangs as done deals to which no changes are allowed
except in dire emergency.  Once you have released a conlang to
the public, this is a good occasion to "freeze" it.  I plan to
make the version of Old Albic I am currently working on such a
final version which may not be changed further except for adding
more words and of course writing more texts.  I have found my
lamatyáve, and the daughters and other relatives of Old Albic
will give me enough work and fun to keep me going!

--
... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html
"Bêsel asa Éam, a Éam atha cvanthal a cvanth atha Éamal." - SiM 1:1