Re: my proposals for a philosophical language

Andrew Nowicki writes:

[snip]

Okay, I haven't been following this discussion all that closely -
hopefully I won't repeat too much that hasn't been said already.

First off, I agree with you that a self-segregating morphology is a
good idea.  At the very least, it greatly simplifies the problem of
developing speech recognition software.  I'm prepared to believe it'll
make language acquisition somewhat easier, and it should prevent
certain ambiguities in speech, too.  These are all fairly minor
concerns (well, speed of learning and ambiguity are important, but the
effect of self-segregation on them is small) but they do exist, so if
you're trying to make a language as useful as possible for everyday
use today, self-segregation is desirable.  All things being equal,
that is.  If it prevented you from doing something more important
efficiently, then it would probably not be worth it.

Fortunately it's not all that difficult to design a morphology which
is unambiguous as to morpheme boundaries.  But not the way you're
doing it.  Your basic problem is that you've restricted yourself to CV
and CVV roots.  Your reasons for doing so are individually sound, but
collectively they've led you to an essentially unworkable solution.
That is _not_ enough roots.

 >
 > We agree that consonant clusters are not
 > desirable.

This as a statement is fair enough.  However, I would argue that
relaxing this restriction would be desirable, as a means of creating
more roots.  Particularly, it would make it easier to create
multisyllabic roots without sacrificing morphemic segregation.

This need not make your language particularly difficult to pronounce.
Very few languages prohibit consonant clusters entirely.  You must
recognize that all clusters are not created equal.  Consider the
following words:

_vprtskvni_ (Georgian, "I peel it")
_ambu_ (Japanese, "saddle")

Now, almost no-one who isn't Georgian could pronounce the cluster in
the first word correctly.  Almost anyone in the world could manage the
second - more, I suspect, than could handle the Ygyde vowel system*.
An intervocalic prenasalized stop is a pretty common cluster.

Now, if you were to allow such clusters, you could put them between
the vowels of morpheme with more than one syllable.  On encountering
the cluster, you know that the morpheme has another syllable to go.

Even if you're determined to restrict yourself to CV _syllables_,
you don't have to restrict yourself to CV _roots_.  Just take certain
vowels or consonants, and say that they can't be used in monosyllabic
roots, but only in the first (or second, or however you want to work
it) syllable of a morpheme.  This will reduce the number of your
monosyllabic roots, of course, but it will increase the number of
roots you can create overall.

Here are some links you may find useful -

http://www.srv.net/~ram/essays.html
        Rick Morneau's essays on language design.  You really should
        read these if you haven't already done so - there's some
        information on creating a self-segregating morphology, and in
        _Lexical Semantics_ a complete language is described which
        might well interest you.

http://www.langmaker.com/dublexnav.htm
        Jeffrey Henning's dublex was an attempt to find the most
        useful morphemes for forming compounds by having people send
        in compounds made from a supplied list of 400 roots.



* Does anyone have any statistics on this?

Reply