phonetic resemblances

[H. Mark Hubey]

----------------------------Original message----------------------------
Sally Thomason wrote:
>
> ----------------------------Original message----------------------------
>
>    Larry Trask and Rich Alderson and a few others are doing a
> great job of explaining why phonetic resemblances are not considered
> by historical linguists to be either necessary or sufficient to
> establish genetic relationships, but so far not much emphasis has

I, of course, disagree completely. It is based on nothing more than
phonetic and semantic "resemblance". In order not to have useless
arguments, I will use the technical term "distance" from now on.

Instead of writing things everyone misunderstands, I will instead
wait for someone to show the genetic relationships of two language
via some examples, and then I will demonstrate why that is about
nothing more than phonetic and semantic distance.


> including words like "animal", which is as close to its French (or
> Latin) source as in Hubey's ironic same-language examples.  In addition,

Excuse me. That is not irony. It is common place in mathematics. A
general
statement has to hold. A person is obviously genetically and familially
related to himself/herself. With a slight perturbation of the same
language example, you can see immediately that dialects are based on
small phonetic distances, family groupings on larger phonetic distances,
and so on. It is trivial to demonstrate that it is so.

> though, you get the very strong phonetic resemblances in mixed
> languages, which for most historical linguists (I think) don't meet
> the standard criteria for genetic relatedness.  It's only where
> both lexicon and grammar show systematic correspondences throughout
> that the languages under comparison fit the model of descent with
> modification from a single parent language.

That is immaterial. The fact is that certain words have been struck
from the list. I posted the constraints before;

1. There must be a "matching" of M words
2. These "matches" must come from a specific set {a,b,c.....}
3. Some words that are close phonetically to the set {A,B,C,...} are
void
4 The "matching" words must match semantically within a tolerance of x
units.
5. The matching words must match phonologically within a tolerance of y
units.

#1. Nobody knows what M has to be. If anyone knows, please post it.

#2. The specific set is usually Swadesh's list, because it is a
formalization
of the ideas inherent in this concept. The general version of it is
that,
some words are technological words and should not count, therefore the
words
must come from some restricted set. There is general agreement on this.

#3. This is the infant-babble argument and it is wrong. I will explain
why
later.

#4. This is the part in which there is confusion presently. I am more
than
happy to show why in every case which any linguist claims that the
concepts of
phonetic and semantic distance IS NOT being used, it is in fact, exactly
that
which is in use and nothing else.

#5. This is the part that comparative linguistics still leaves flaky and
in which
it it easy to show that the same concept is used.


>    Compare English with Tok Pisin, for instance, and you'll see
> very strong phonetic resemblances throughout the vocabularies, but

That is rule number 2. The fact is that the specific set of words is
not the ones. IT says nothing about the fact that phonetic distance is
not being used. It is being used, but another rule determines that some
words cannot be used.

> etc.  Only a systematic comparison of vocabulary *and grammar* will

1. systematic comparison has to allow for phonetic distances of words
which are said to be technological and hence should be disallowed so
it is not systematic phonetic distances that are used, only the phonetic
distances from a specific set of words.

2. The grammar part is trickier. I would leave it for later because it
is
much more difficult than vocabulary. I don't mean that I want to
disallow it.
No, it is not really much more difficult than the other. But we should
probably do things one step at a time.

> exception, and there even the most fervent believer in the
> resemblance-is-all-you-need school will realize that to classify it
> genetically you'd have to toss a coin to decide whether nouns or verbs
> win.

You are using "phonetic resemblance" in a way that is not clear. I think
what you want to say is something else because I do not see it the same
way and I
tried to show how the concept of resemblance and similarity is related
to distance and how it is used in other fields, and even made it
rigourously mathematical. YOu mean to say something else, but it is not
clear to me. (It really is, but maybe not
clear to others :-)).

I have listed the general rules which are in use, even if they are
foggy.


PS. The concepts "systematic", "resemblance", "grammar" etc have usages
which
can (and does) lead to great problems. The only solution that has been
found
in the rigourous sciences it to create precise definitions. I did so
with phonetic
distance. An analogical one can be made in semantics, and should be
made, but there are no takers.
--
Best Regards,
Mark
-==-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
hubeyh at montclair.edu =-=-=-= http://www.csam.montclair.edu/~hubey
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=