Re: General Rule

Richard Wordingham wrote:

--- In Nostratica@yahoogroups.com, "H.M. Hubey" <hubeyh@...> wrote: > > > Richard Wordingham wrote: > > > --- In Nostratica@yahoogroups.com, "H.M. Hubey" <hubeyh@...> wrote: > > > > > > > > > Richard Wordingham wrote: > > > > > > > --- In Nostratica@yahoogroups.com, > > > > "H.M. Hubey" <hubeyh@...> wrote: > In general I do not believe in the split into language families first > and then > further splits into subfamilies etc. Specifically, I said that if there > is Nostratic > and if Nostratic is the common ancestor of IE, AA, Turkic etc. then if the > family splits occurred, we still have to start from the common ancestor and > split into families. The principle is that Turkic evidence has to be taken > into consideration. > > Now, the way things were produced although "body parts" were used > in the Swadesh list etc, only those body parts that made the IE family > look good were used. For example, Turkic has auz (mouth), Hittite ais, > Latin o:s. Turkish has omuz (shoulder), etc. The point is we still have > to take into account Turkic. Bomhard's excellent book is very light > in Turkic mainly because Turkologists use a Zeroth Postulate more than > anything else: Mongoloid, Mongoloid, Mongoloid. The data has been > twisted out of shape, and if Turkic is taken into account there could > be serious changes. Turkic has words that are so archaic, it can be > used to reconstruct Akkadian. I see now that it had to be this way. > They got lost in the steppes for a long time, then mixed with some > Mongoloids > and historians, especially led by the xUSSR, have been busy keeping the > Eurasian steppes clean and squeaky for Aryans, misled by some historians > who deliberately falsified the records. > Here is a simple example. Suppose we use *patar for father. Even nonAltaist > Turkologists agree that the prototurkic initial-p changed to a bilabial > fricative, > like in Japanese, and then to h-, and then got lost in most dialects except > Khaladj. Turkic also seems to have lost most final liquids. So we can easily > get ata from *patar. That means there is no reason to avoid the ata, > ama, mama > words and that immediately throws a monkey wrench into the works. Even today > there is no agreement, after 200+ years, and this will make it even > worse. It is > easy to see that athir went through similar changes as ata. The immediate problem here is that we are talking about 'nursery' words. The probability that a language will have such a word is fairly high (typically 20%?), and that has to be remembered when using such words to make comparisons. What makes the words useful as an indicator of IE-relatedness is the suffix-like '-ter' (currently seen as PIE *-h2ter).
That suffix, in more general formation -thar occurs in Turkic. And nursery words would be of
more interest for Nostratic precisely because they may truly be the words least resistant to
change. And they will probably show up in other books.

Nursery words in other languages, and onomotapaic words in various languages are not
exactly like each other. Child's speech differs from language lto nguage.

> Let me continue with this example. In this case *matar does not work, > because > Turkic (and Hittite) has ana, amma, etc. So then we might try *apatar, and > *amatar. Then it looks like apatar is a duplicated word e.g. apa- ata. > But both of > these words mean father in many languages. Therefore it looks like this > duplicated > word is created from an earlier split. Why a split? English has both 'pa' and 'daddy'.

There has to be some general rules of sound change. There has to be some directionality
if mathematics is to be used. To me a rule is p > t > k, because a bilabial is the first
consonant an infant learns. So to me ata and apa are roots from an earlier language with the
change. All of this cannot be via two heuristics of linguistics. We also have archaeological
evidence and genetic evidence.

Before anything else, I think models should be made clear and there should be some
effort at computation and the use of some optimization principle like in use in other fields.
AT least then models can be compared to each other.

> But then we also get, as a bonus, > ama, Latin amore, Turkic amrak (loving), imren (to covet), em (to suck), > emchek (breast), am (vagina), yum (to close, eg. mouth?). The neat split > into families does not look so neat anymore. How do you cite Latin nouns? Latin amo:re 'love' is ablative singular, whereas I would expect to see the nominative (amor) or accusative singular (amo:rem) cited. What is 'ama'? The stem of the Latin verb to love? Mind you, citing the infinitive ('ama:re' in case any lurker is unaware) would give the impression that you were comparing the /r/ of the Latin suffixes (< /s/) with /r/ in the Turkic words :)

The root in Turkic is Vm; em, am, im but the yum could possibly point to an earlier form. The -re
is a verbal suffix and -en is reflexive. im-re-en.

'Mama' is, worldwide, a very common starting point for words meaning 'mother' or 'nipple'. (In our family, our first baby gave us the word 'maman' for food, which my wife tells me is Chinese!)

Yes, and it could be the oldest such word.

> There is plenty of evidence that Turkic even retains words form > pre-Semitic,e.g. > more archaic words than those that exist in Akkadian. It has too many words > from the Mideast Sumerian, and Akkadian. Loans or common inheritance?
How can you tell the difference except via a global optimization? At least the rules would be
clear if nothing else and there would be a basis for comparison at least of the various efforts
of IEanists to reconstruct IE. Right now, after 200+ years even that does not exist.

> > The point is that words on the Swadesh list do get replaced, > > sometimes by internal development, sometimes by borrowing. > > > Yes, and which words are used should be used iteratively e.g. select the > set N and look > for a family. If the family does not look good, change the set and look > at the family. That approach throws away evidence! It must complicate the statistics enormously!

Lacking some standard test, everyone just keeps creating things. Right now I want to create
math models, not statistical models. Let Kessler and others continue their way.

The simplest thing would be to use a Swadesh-like list and a Kessler-like test, one that
can be used as a "distance" between two languages, and one that does not depend on
independence.

> So the original set was selected deliberately to make a neat family. And > that is > where Kessler's book is extremely important. The list must be selected > independently > of the family. For establishing a family, yes. > If body parts are to be used, then all the parts should > be used not only > those that make Latin, Sanskrit, Greek belong most tightly or something > similar. I think > it is really time to look at the list much more carefully, this time > more carefully > making sure that independence is preserved. Some parts bind their referring words better than others. An alternative is to look at data from several families, and select items that way. There is a very large database for Austronesian.
Yes, and this is all nice, and all heuristics. Now we have hard numbers.And that is what
Kessler has done.

> > > The > > history of the Swadesh word lists is interesting; they started out as > > meanings for which every language should have a word. A few meanings > > on the 100-word list are particular liable to replacement - 'road' is > > one example, and some much stabler meanings are omitted from the 100- > > word list - e.g. 'salt', 'snow'. 'Snow' is missing because it is not > > an important word in hot climates. I am not sure about 'salt'; > > perhaps it is a less stable among Australian Aborigines, who have > > less use for it than the rest of us. > > Time to redo. That is especially pertinent for Nostratic. > "road" should show up after use of animals as draft animals, more or less. > > I think more words should be used, and the words should be given weights > so that we > can estimate things about what time periods the words started traveling. Weighting is one thing that calculated replacement rates can be used for. Mind you, I got criticised for using IE replacement rates to assess a comparison of Austronesian and Austro-Asiatic (or rather Malayo-Polynesian and Mon-Khmer as the data turned out).

Even weighting requires a model.

I think it would be desirable to find some way of capturing cognates when their meaning has drifted apart. It looks laborious, if only because some way would have to be found to compensate for allowing more false matches in.

Yes, it would be.

In any case I can send my papers to anyone who asks for them.

Richard.
To unsubscribe from this group, send an email to: Nostratica-unsubscribe@yahoogroups.com

Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.

-- M. Hubey -o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o The only difference between humans and machines is that humans can be created by unskilled labor. Arthur C. Clarke /\/\/\/\//\/\/\/\/\/\/ http://www.csam.montclair.edu/~hubey