Morpheme Self-Segementation Strategy

[Dirk Elzinga <DELZINGA@CCIT.ARIZONA.EDU>]

Hello, CONLANGers.

On Sat, 11.Feb 1995, Bob Michael wrote (among other things):

>I chose to have morphemes end with vowels to avoid unexpected
>or unpronounceable consonant clusters.  I chose a, i, and u for initial
>vowels because they are more differentiated from each other and would
>normally be stressed.

This is a very interesting idea.  It is essentially the same idea that 
Roman Jakobson had in his essay Child Language, Aphasia, and Phonological
Universals (if I remember the title correctly) with one important
difference.  Jakobson's proposal was that languages will create as much
space as possible between elements of a system (in this case, vowels)
to maximize the salient distinctions that need to be made.  So if a
language has only three vowels, they will most likely be {i,a,u}, since
this is the set with the most space around each member.  He also makes
the claim that _unstressed_ vowels in a language will also tend to be
maximally distinct in the same way since stress could not be used to
give prominence to the vowels.  Suppose that a language has five vowels:
{i,e,a,o,u} in stressed syllables.  Jakobson's assertion then, is that
if any reduction occurs in the vowel system for unstressed syllables, this
reduction will eliminate {e,o} and keep the vowels that are maximally
distinct from each other, since other means of prominence have been
denied them.

Of course, there are bound to be counter-examples in the world's (natural)
languages, but I think the idea is sound.  As part of a scheme or
strategy to make word-boundaries in a constructed language clear, I'm not
sure that it's really important what vowel qualities are used, only that
the distribution (stressed syllable, unstressed syllable) of the vowels
is tightly controlled.  Of course, if the scheme is grounded in some
sort of (perceived) phonetic reality, it just makes it that much more
satisfying to the language creator and the language learner.

Dirk Elzinga