Re: Arabic and BACK TO Self-segregating morphology

On 12/21/05, Gary Shannon <fiziwig@...> wrote:

> This is a really interesting problem, from the
> theoretical viewpoint. But as a practical matter, I
> wouldn't want any conlang of mine to have words as
> long as "kainalijaupalitaosa" anyway, so I doubt I
> could find any real-world use for anything more than
> the most rudamentary system of compounding. In the
> Brown Corpus, the longest word by syllable count is
> "individual" at five syllables, and even many of the
So, if your system requires all lexical roots to be
three consonants, and all derivational vowel patterns
to be three vowels, you would probably want to allow
a limited amount of derivational prefixing and suffixing,
but no compound words as such?  That still
leaves a few ambiguous cases, such as where
a word has a prefix and a suffix, and the order
of operations is not clear between

(prefix + word) + suffix
or
prefix + (word + suffix)

There are words like this in Esperanto,
but for most of them the two possible
parsings are semantically as near
identical as makes no difference.  E.g.,

mal-sagx-ul-o

It doesn't matter if we parse it as "the opposite
of a wise person" or "a person who is the
opposite of wise".

If you allow some consonant clusters, then
you can have some vowel patterns with
null spaces  that allow two consonants of
the roots to adjoin, e.g.

NLJ + e-_-u => nelju

- and if those were the most frequently occurring
vowel patterns, then you could have many compounds
that aren't excessively long.  You might also have
some vowel patterns that have an optional
middle vowel, used only if a forbidden cluster
would result from its omission; e.g.,

NLJ + e-(i)-u => nelju,
but
PJJ + e-(i)-u => pejiju

In another message, you wrote:

> SUMMARY: I'm looking for a systematic way to generate
> a lexicon. Not the actually letters and syllables of
> each word, but the language-independant concepts and
> meanings, and how they relate to each other. I'm
> looking for some systematic enumeration of all the
> different ways in which two closely related words can
> actually be related to each other. In other words, a
> list of formal definitions for the relationships
> between the members of pairs like "to teach:teacher",
> "to learn:to teach", "beach:sand", "to break:broken",
> etc. I've come up with 9 pairs of relationships (18
> relationships in total) so far, but I'm sure there are
> a LOT more.
We talked about this here a while ago, and I don't think
anyone else offered a comprehensive list of the
kind you're looking for.  I would be interested in
such a thing as well; possibly we could collaborate
on it through e.g. the Conlang Wikicity?

--
Jim Henry
http://www.pobox.com/~jimhenry

Reply