Re: Arabic and BACK TO Self-segregating morphology

--- Patrick Littell <puchitao@...> wrote:

> On 12/19/05, Jim Henry <jimhenry1973@...>
> wrote:
> >
> >
> >
> > If compounds of two or more words are allowed,
> > and your derivational patterns include both
> prefixes
> > and suffixes, it seems that a word of seven
> syllables
> > might be ambiguous, e.g.:
Word consists of three consonants, each followed by a
single vowel, or followed by a single vowel followed
by 'N' (which is excluded from the list of valid
consonants). Prefix consists of consonant followed by
vowel pair ending in "o" or "u". Suffix consists of
consonant followed by vowel pair ending in "i" or "e".

> > (prefix + word1) + word2
kiotakamintosanti = CVoCVCVCVnCVCVnCV, for example, is
unambiguous.

> > or
> > (word1 + suffix) + word2
lokinsasaitataban = CVCVnCVCViCVCVCVn, for example, is
unambiguous.

> > or
> > word1 + (prefix + word2)
CVnCVCVnCVuCVCVnCV, for example, is unambiguous.

> > or
> > word1 + (word2 + suffix)
> >
CVnCVCVCVnCVnCVCVi, for example, is unambiguous.

> > This might also be an ambiguity re:
> > the word boundary in a two-word phrase.
word boundries might be ambiguous, but context should
disambiguate them, and if it doesn't then the
consecutive words would have the same meaning as the
same words compounded anyway. So no information is
lost.

> Well, if we're requiring that it not just have
> self-segmenting morphemes,
> but also be structurally unambiguous, we've got some
> serious work cut out.
> Gotta disambiguate these, too:
>
> prefix + (word1 + word2)
CVoCVCVnCVnCVCVCV, for example, is unambiguous.

> (word1 + word2) + suffix
CVCVCVCVCVCVCVi, for example, is unambiguous.

> And once you get more than three... well, it would
> be a nightmare to work
> out completely unless serious restrictions were put
> on word-formation.
The restrictions above are not that serious, and they
get the job done.

> Hmm, that's an engineering problem to give one
> pause.  Not just word
> segmentation or morpheme segmentation or both of
> these at once, but both of
> these *plus* unambiguous morphological structure.
>
> Any takers?
Been there, done that.

In fact here's another scheme: words are CV(n)...CV(n)
as above, but not restricted to three consonants.
Prefixes and suffixes also have the form CV or CVn.
Prefixes are attached with a "linking: 'i', and
suffixes are attached with a linking "o". When words
are joined a linking "u" is used. Thus:
(prefix)i(word)o(suffix)u(word).

[ka+(i)+sita+(o)+so]+(u)+minka = kaisitaosouminka
which is unambiguous, and probably means exactly the
same thing as: kaisitaoso minka anyway.

Just think in terms of tinker toy modules made of
green and blue sticks, assembled into a large
structure by using only red sticks to form the
connections. No matter how comoplex the final
assembly, the component subassemblies can be
rediscovered by simply removing all the red sticks. In
this case, substituting spaces for I's O's and U's
that follow a vowel or an 'N'.

--gary

> -- Pat
>

Reply