Re: VMs: RE: zara

[Koontz John E <John.Koontz@xxxxxxxxxxxx>]

On Tue, 26 Oct 2004, Marke Fincher wrote:
> With the constraints of grammar in a meaningful language there will
> be a reduced set of words that can validly follow a given word,
> but knowing just the *ending* of the previous word wouldn't normally
> constrain the characters that begin the next word, would it?

In some languages here are interword assimilatory processes that constrain
the phonology of following words in certain word sequences or grammatical
constructions.  This is characteristic of Celtic.  Similar things occur
elsewhere around the world, e.g., in Fulani (Africa) and Comanche (North
America) (and among their respective relatives).  Tolkien - inspired by
the Celtic data - had a lot of fun with this in deriving Sindarin from
Quenya.  Maybe too much fun.

To use a purely invented example, here are the sort of things that can
occur:

- final nasals (n or m) cause nasalization of a following word initial
consonant, e.g., an ba => a ma

- final vowels cause fricativization of a following word initial
consonant, e.g., a ba => a va

- final stop consonants yield geminates in following words, e.g., at pa =>
a ppa

If the final elements producing the changes are lost, but the effects on
following elements endure, the system can be grammaticalized so that words
have varying initial sounds depending on what precedes.  A lot depends on
whether forms like an and at continue to exist in other contexts.  If not,
then you end up with three forms "a" that have different effects on what
follows.

However, as I understand your observation the final element of the
preceding word is what conditions the initial element of the following
word - they are correlated - and that is a bit different, more like an
intermediate state of the above in which an ma, a va, and at ppa are
written (whatever the actually phonology).  And, of course, as you have
suggested, it could arise from inserting chunks in which the spaces are
chunk-internal.

Historically, spaces are not always used to separate words in writing, so
perhaps we should treat space as a character.  Alternatively something
other than a space might be used as a separator, e.g., in some historical
systems a raised dot.  If spaces are arbitrary or misleading in the VMs,
might some other character be more plausible in that light?

Another source correlation that could arise would be gender marking.  For
example, if nouns ending in a given gender marker are followed by verbs
beginning with a corresponding subject-gender marker.  I don't have a
specific example in mind.  The cases I know of where third person gender
marking occurs in verbs and verbs involve prefixal noun and verb marking,
e.g., in Niger-Congo languages, or suffixal noun and verb marking, e.g.,
Afroasiatic.  In such cases you don't get the gender markings adjacent.

In any event, both the "natural" explanations I've offered would only
apply across certain word boundaries, not all word boundaries.  Do you
have some statistics for cross space-correlations, or is this still
impressionistic?

______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list