Re: Can realism be retro-fitted?

Sorry for late answer.  Real Life and so on...

Herman Miller skrev:
 > I've been wondering about the possibility of taking some
 > of my existing languages and developing their historical
 > background to make them more realistic and less artificial-
 > seeming. I recently finished reading Guy Deutscher's _The
 > Unfolding of Language_ which has been mentioned here
 > lately; one of the things that stood out was the role of
 > analogy in creating new patterns.

Analogy is important, but it should not be overdone or
overstated, especially not in phonology. If it were all-
persuasive comparative reconstruction would be impossible,
which it probably isn't -- even though cases where the
ancestral language is known, like Romance, show that reality
is rather complicated, and that reconstructed protolanguages
are highly idealized. However few today would go so far as
to deny the regularity of sound change as Schuchardt
<http://en.wikipedia.org/wiki/Hugo_Schuchardt> did. If you
read German the whole story is at:
<http://tinyurl.com/2emh6j>. The WP article on the
comparative method takes up most of the issues too:
<http://en.wikipedia.org/wiki/Comparative_method>.

In spite of all this the assumption of exeptionless sound
laws is a methodological necessity for doing reconstruction,
since it is the only thing that allows us to project
anything backwards at all, and experience from Romance and
other similar cases show that it is correct very much of the
time -- over 90% usually. In fact we have to assume
regularity until actual evidence points another way, since
else the line between evidence-based reconstruction and
arbitrary a-priori conlanging would be erased. (This is not
to denigrate arbitrary a-priori conlanging, but if you are
into both conlanging and comparative linguistic
reconstruction you need to keep in mind which hat you're
wearing at any moment! :-)

 > I'm familiar in a general sense with how the sounds and
 > meanings of words can change over time. But one of the
 > problems I've always had in reconstructing the history of
 > a language that doesn't have one is trying to come up with
 > a consistent set of rules that relate the sounds of the
 > hypothetical older language to the already documented
 > language.

I would recommend to decide a-priori on the lámatyáve --
phoneme inventory, phonotactics -- of the ancestor, compare
it to that of the descendant and brainstorm for ideas how
the phonemes of the descendant may derive from the phonemes,
allophones and phoneme combinations of the ancestor. Even
though real language change isn't telic it has its
advantages to work telically in historical conlanging! :-)
In fact I think it's necessary, unless you devise a computer
program which randomly chooses and applies soundchanges --
which of course would be an interesting thing to do! :-q

 > The problem with taking an existing language to start with
 > is that I have to come up with a historical explanation
 > for each feature of the language, or modify it in such a
 > way that I can more easily explain it. For instance, when
 > I examined the tone patterns of two-syllable words in
 > Simîk, I noticed that a few patterns were much more
 > comman than the others, which could be explained by
 > development from a simpler tone systme in earlier versions
 > of the language. But not all tone patterns fit into that
 > system, so I had to assume they were borrowed from some
 > other language, or were different in some other way
 > (having a special tone pattern that was used for
 > emphasis).
 >
 > In the long run, is it better to start with one or more
 > artificial proto-languages and develop them forward
 > through time

That's the way I usually do it.

 > (which involves a lot of work on features that may not
 > even make it into the future language system),

Not necessarily; since you are in a position to decide what
features you want in the descendant language you need design
only those features for the protolanguage. The analog in
real historical reconstruction is that no features that have
been lost in all descendants can be reconstructed for the
protolanguage.

This makes a reconstructed protolanguage a rather special
thing, and different from the actual prehistoric language(s)
that once existed in that you can only reconstruct the
regularities and those parts of structure which survive --
or leave a mark, the technical term is "leave a reflex" or
"be reflected" -- in the descendant languages. Any
irregularities and anything which analogy, phonetic loss
('merger with zero'), syntactic and morphotactic change
(which AFAICT are forms of analogy -- spontaneous or
deliberate innovation by speakers is probably possible but
even less reconstructible than irregularity) has done away
with cannot be reconstructed, so a protolanguage (
'*asterisk language' ) is only a subset of the actual
prehistoric language. Interestingly Tolkien implicitly
raised some of these issues -- notably deliberate innovation
by speakers -- in his historical conlanging.

However the real fun begins when you develop a protolanguage
with certain phonemics, phonotactics and grammatical
features, then design different historical phonologies(*)
leading to different descendants, and see what kind of havoc
the different historical phonologies wreak with the
phonemics, phonotactics and grammar. That is IMNSHO the real
gratification of diachronic conlanging.

(* I'm in the habit of speaking of "(synchronic) phonemics"
and "(historical) phonology" -- this is because in
comparative philology the unqualified word "phonology"
always has an historical sense, which has rubbed off on me.
I know that in a restricted sense "phonemics" is a specific
subbranch, but my usage has the advantage of preserving the
syn-/dia-chronic distinction even when I slip and forget the
adjective "historical", which I often do.)

 > or to start with an existing language and develop a
 > history for it?

I tried that with Sohlob, but the protolanguage (Kijeb)
ended up somewhat bland and as I made the phonological
changes more and more interesting the descendant language
ended up changing almost as much as the protolanguage. At
least for me this is a Good Thing and not something to
resist, since it enhances naturalism in both ends. I regard
it as a natural part of a-priori historical conlanging. The
more organic it is the better!

 > For a specific example, I thought of taking Tirelat and
 > trying to develop a history for it. Tirelat is a very
 > regular and artificial language, which may actually be a
 > result of engineering a more natural language to eliminate
 > irregularities. But to start with something simple, the
 > vowel system: Tirelat has 7 vowels /a e i @ 1 o u/, which
 > may be long or short. Diphthongs do exist, but only /ai/
 > is common; combinations like /ia/ and /ui/ can be analyzed
 > as consonant + vowel (/ja/, /wi/), except for the fact
 > that there is no /ji/ or /wu/. So where do these 7 vowels
 > come from?

The absence of /ji/ or /wu/ is IMO not a problem at all --
it is just a realistic phonotactic constraint -- I must
admit that my Kijeb tolerates none of /ji ij wu uw/; any
instances that arise in morphology turn the /j/ or /w/ into
/g_j/ or /g_w/. Mandarin has the opposite constraint with no
/i u y/ word initially -- [ji Hy wu] and [i y u] are in
complementary distribution, and in fact Bopomofo doesn't
distinguish them, but has a single grapheme for each pair.
<digression> Writing /ly/ and /ny/ as _lyu_ and _nyu_ would
be perfectly possible without other changes of Pinyin, and
would do away with the need for _ü_.</digression>

 > One thought that might explain the long vowels is that
 > earlier versions of the language had more diphthongs,
 > which simplified to single vowels. Notably, /a i u/ are
 > more common than the other vowels.

Of course /a i u/ is a perfectly possible vowel system. May
the other vowels have arisen as positional allophones? In
Kijeb i let umlaut phenomena transform a 3-vowel system into
an 9-vowel system, which then shrinks to 6 or 5 vowels
through mergers -- the merger patterns being different in
different dialects -- and the rise of vowel harmony, which
then in one dialect turns again into an 8-vowel system
through loss of intervocalic /j w G h/ and
monophthongization of the resulting diphthongs which gives
rise to front rounded vowels which did not exist in the 6-
vowel system. Phew, got all that ;-)

Looking at the Tirelat script page you referred to
<http://www.io.com/~hmiller/lang/Tirelat/script.html> (there
alas no table, but I sketched one to get an overview) i see
that one way you could go would be to assuume an earlier
phonemic system like:

   *p *p;  *t *t;    *tS *tS;  *k *k;

   *b *b;  *d *d;    *dZ *dZ;  *g *g;

   *f *f;  *s *s;    *S *S;     *x *x;    (*?)

   *v *v;  *z *z;    *Z *Z;     *G (*G;)

   *m *m;  *n *n;               *N *N;

   *w (*w;)             *j                (*h)

           *r *r;

           *4 *4;

           *l *l;

Items in parentheses may or may not exist: *w; and *G; may
be non-distinct from *j.  *N; is probably not distinct from
*n; and the single phoneme may be [J].  *? and *h may come
in handy to get diphthongs later! :-)

*t and *d have [T D] allophones from the outset.  Their
palatalized versions may be nondistinct from *s; *z;.

And the vowel system: *a *e *i *o *u --

Each of the vowels has different allophones after
palatalized and unpalatalized consonants like so:

           V         ;V

     *a    A         &

     *e    E         e

     *i    (i\)      i

     *o    O         3\

     *u    u         y


Sound changes:

Palatalization is lost as a feature in consonants, but with
some odd mergers and splits:

*j > 0 (zero) word initially and perhaps between vowels.

*tS *dZ  > ts dz

*Z, *Z;  > j (distinct from *j)
     -- this would explain the existence of /j/ before non-
     palatalized vowels.  *G; is another possible source of
     /j/ if /G/ isn't too frequent before palatalized vowels.

*S, *S; > s` -- uncontroversial?

*lh/*hl > K
*rh/hr  > r_0
     -- /K/ may also arise from clusters of sibilant and *l.
     Maybe *s(;) l(;) (*z(;)l(;)) > /K/ while *t(;)l(;) ^d(;)
     l(;) > sl zl -- it all depends on what clusters actually
     appear in Tirelat. If /j/ doesn't appear before /l/ or
     /L/ then *Z(;)l(;) may be a source too.  Similar for
     /r_0/.

*r / r(;) > z`

*h > 0 (zero)
*? > O -- at least non-word-initially.

*[A], *[&] > *a
     -- this actually happened in the history of Northern
     English.  Thus the difference between *;a and *a is
     ineffectual.

*[y] may go with *[i] or *[y] > u\ > i\. A possible source
of dialect difference.  It would depend on the relative
frequency of /i/ and /i\/ in Tirelat.

*[E] merges with *[3\] as /@/, then *[e] > /E/.

The vowels in hiatus resultant from the loss of *j *h *?
are resolved by becoming various diphthongs and long vowels.
In Mongolian it is, interestingly, the quality of the second
vowel in *VhV and *VGV clusters which determines the quality
of the resultant long vowels.  Possibly *ai > /a:/ and *au >
/O:/ while new /ai/ and /au/ arise from *ahi, *ahe *aje,
*ahu, *aho etc., while *eha, *oha etc. give /E:/, /O:/, or
/a:/ depending on the relative frequency of these vowels.

Admittedly this doesn't take you all that far back, but it is
a start.

 > But any patterns in the data would have to be coincidental
 > at this point, since the Tirelat vowel system wasn't
 > developed with a history in mind. So in a case like this,
 > would it be better to come up with an arbitrary history
 > that doesn't fit all the facts (e.g. proto-language /o/
 > develops into /@/ except in the vicinity of a bilabial
 > consonant, where it remains /o/, after which other
 > phonetic changes occurred which caused /f/ to merge with
 > /x/ resulting in a phonemic distinction between /x@/ and
 > /xo/)? As it happens, /xo/ is slightly more common than
 > might be expected, but /p@/ and /b@/ do exist, which can't
 > be explained by the hypothetical sound changes.

I hope the model I proposed above would resove this: /p@/
and /b@/ are from *p;o and *b;o, possibly also from *pe and
*be distinct from *p;e and *b;e which give /pE/ and /bE/.

Later you wrote:

 > Much of the Tirelat vocabulary is in a disorganized state
 > (from all the changes that happened to the language;
 > unfortunately I got carried away with the search-and-replace
 > feature), but that could be an advantage since if one
 > explanation doesn't work in one case, an alternative form of
 > the same word may exist. One problem is that I developed the
 > vocabulary for years as a language that could be used by
 > humans on present-day Earth, but later assigned it to the
 > non-human Sangari on a different planet. Still, even though
 > I need to assign new meanings to many of the words, I should
 > be able to find local equivalents for many of them.

Yes that may be a good hunting ground both for designing
history and dialects.  It also means that you are in a
position where it may hurt less to change the modern language
as occasioned by the development of the history.

I see now that you may also start from an earlier
seven-vowel system:

     *i                       *u

         *e               *o

             *E       *O

                 *a

with a following somewhat skewed chain-shift:

*e > *E
*E > & > a
*a > Q > O
*O > o > @
*o > u
*u > i\

o > @ happens when there are only three heights in
the front vowels, to remove one height in the back,
assuming that /E/ /O/ are really mean mid.

As you see there are several possibilities! :-)
--

/BP 8^)
--
   B.Philip Jonsson mailto:melrochX@melroch.se (delete X)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Truth, Sir, is a cow which will give [skeptics] no more milk,
and so they are gone to milk the bull."
                                     -- Sam. Johnson (no rel. ;)

Replies