Re: Optional Soundlaws

At 7:23:16 PM on Monday, October 25, 2010, stlatos wrote:

> --- In cybalist@yahoogroups.com, johnvertical@... wrote:

>> --- In cybalist@yahoogroups.com, "stlatos" wrote:

>>> I don't know why this seems so hard for some people to
>>> understand. A change in a sound is no less of a law if
>>> it has two outcomes. For example, n > l (opt.) is the
>>> same as a law n > l OR n > n

>> These are correspondences. Laws, by definition, don't
>> have "or"s. You can call your components n > l and n > n
>> laws individually, but not their disjunction.

> Wrong.

He is correct as the term is usually used and understood.
You can reasonably speak of a sound law '/n/ is retained'
even if there are sporadic instances of /n/ > /l/, and you
can reasonably speak of a sound law '/n/ > /l/' even if
there are sporadic retentions; but if instances of retention
and of /n/ > /l/ are roughly equinumerous and have (as yet,
at least) no conditioning factor, you haven't (in the usual
sense of the word) a law at all. At best you've a
description of a relatively weak sort of regularity in the
data; at worst you've a deus ex linguista.

This is not just an argument about the meaning of the word
'law': the more a proposed derivation relies on
unconditioned, apparently random variation, the shakier and
less persuasive it necessarily is.

>>> (analogous to 2 or -2 being the square root of 4).

>> And "square root" in this sense isn't a function, so a
>> fair analogy. (Tho a square root isn't "sometimes 2,
>> sometimes -2", but "2 and -2 simultaneously".)

> Both the sounds n and l exist simultaneously within the
> system of a language; any instance of pronunciation is
> sometimes n, sometimes l, so the analogy seems fine
> (though it needn't be that precise in correspondence for
> my purposes).

As John already clearly explained, it isn't fine: 'any
instance of the square root of 4 is sometimes 2, sometimes
-2' is a nonsensical statement, while 'any instance of
pronunciation of the phoneme is sometimes [n], sometimes
[l]' is not.

>>> Many of these changes are known. Instead of
>>> criticizing my methods, learn about what is already
>>> known. For example, in Salishan, n and l alternate.
>>> There is no regularity, no dialect mixing, only
>>> optionality. In a loanwoard like school > skun, it's
>>> easily seen by linguists, the people who speak the
>>> language know about it, there's nothing else to say.
>>> The alt. l/n exists across most of the Americas, and
>>> obviously is either from the parent l. of them all, or
>>> an incredibly old areal change, borrowing, etc. Since
>>> it is also found throughout Asia, nothing else is
>>> likely.

>> It sounds like you are confusing different phenomena.

You're assuming that the facts are as stated. Perhaps they
are, but I've no reason to take Sean's word for it,
especially considering what the last two sentences imply
about his assumptions.

[...]

>> There's nothing simple in optionality as you must specify
>> for each and every applicable word separately whether it
>> undergoes the change or not.

> Wrong. All applicable words underwent all the opt.
> changes.

An idiosyncratic way of saying that some words underwent the
change and others didn't. Worse, a misleading way of saying
so, since it suggests a uniformity of process that need not
be in evidence.

Brian