Newsgroups: sci.lang
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!uunet!munnari.oz.au!comp.vuw.ac.nz!actrix.gen.nz!zohrab_p
From: zohrab_p@atlantis.actrix.gen.nz (Peter Zohrab)
Subject: Re: historical linguistics
Message-ID: <D4EsKK.Byt@actrix.gen.nz>
Sender: news@actrix.gen.nz (News Administrator)
Organization: Actrix - New Zealand Internet Service Provider
Date: Wed, 22 Feb 1995 16:02:42 GMT
References: <D3y5L0.B10@actrix.gen.nz> <3hof5k$o74@darkwing.uoregon.edu> <D402w9.5wG@actrix.gen.nz> <Pine.SOL.3.91.950217131247.10328A-100000@darkwing>
X-Nntp-Posting-Host: atlantis.actrix.gen.nz
Lines: 180

In article <Pine.SOL.3.91.950217131247.10328A-100000@darkwing>,
Scott DeLancey  <delancey@darkwing.uoregon.edu> wrote:
> On Tue, 14 Feb 1995, Peter Zohrab wrote:
> 
> > In article <3hof5k$o74@darkwing.uoregon.edu>,
> > Scott C DeLancey <delancey@darkwing.uoregon.edu> wrote:
> > > In article <D3y5L0.B10@actrix.gen.nz>,
> > > Peter Zohrab <zohrab_p@atlantis.actrix.gen.nz> wrote:
> > > >When a language or language group (say Thai or Tai) resembles one language or
> > > >language-group ( say Chinese) phonologically, morphologically and
> > > >syntactically (aka typologically), but does not resemble it in that portion of
> > > >its vocabulary which is deemed to be least likely to have been borrowed, by
> > > >what precise chain of logical reasoning does one conclude that the two
> > > >languages/language-groups are not genetically related, which is what people do
> > > >seem to conclude -- or by what reasoning do you reach any conclusion at all in
> > > >such cases ?
> > > 
> > > You've not got this quite right, actually.  There are very few
> > > morphological resemblances between Tai and Chinese.  After all,
> > > neither of them have much morphology to resemble.
> > 
> > I'm not sure that that is quite fair.  It is plausible, at least, to say that
> > Chinese and Tai have exactly the same morpheme [0] in a large number of cases
> > where other languages have different morphemes.
> 
> Myself, I don't think it's correct, or even plausible, to say that.
> I'm not crazy about the notion of zero morphemes at all, but at best
> that's only a legitimate notion in the case where the zero is part of
> a paradigm.  So, we can, if we really want to, say that the English present 
> tense morpheme is zero, because it contrasts with the past tense
> morpheme <-ed>.  But you can't say that English has a zero object marker,
> because English doesn't have case markers at all.

That depends on the context.  If you're just analysing English on its own,
then it makes no sense to say that English has a zero object-marker (for
nouns, not pronouns), as it has no case-markers at all for nouns.

But if you're doing a comparative/historical analysis of the potential
relationship between English and some language L, and language L does have an
object marker, then it must be relevant to say that English has a zero
object-marker.

The theory of Historical/Comparative Linguistics seems overly implicit (like
the theory of intuitions in Generative Grammar), and my cynical mind tells me
that, in both cases, this is because being explicit would reveal enormous
problems.

Underlying comparative lexicography, if that is the right term to use, is some
rather simple arithmetic, it seems to me.  Basically, if the word for "dog" in
English and Language L are similar, then this raises the POSSIBILITY that
English and Language L are related.  If the words for "cat" are also similar,
then this increases the probability that they are related, and so on.  The
more similarities, the greater the chance that they are related.

This is complicated by the fact that the words for "dog" etc. in English and
Language L are seldom exactly the same.  An element of judgement has to be
used in deciding whether the differences should be regarded as irrelevant or
not.

Linguists do have their ideas as to plausible sound-changes, but these sound
changes are never inevitable or without exception, so it is impossible to be
sure (say) that an English word beginning with /p/ is related to a German word
beginning with /pf/ (assuming that they are otherwise identical and have
similar meanings), unless there is corroborating evidence.

What I am suggesting is that there should be some explicit metric which gives
relative weightings to individual, purported lexical resemblances, on the one
hand, and morphosyntactical resemblances, on the other -- for the purpose of
establishing genetic relationships between languages. 

>      Anyway, what you're suggesting here is that there's some significance
> to the fact that Chinese and Tai both have, for example, no overt marker
> of subject or object.  But again, this is of minimal value in determining
> relationship, because it's not arbitrary enough.  These languages share that 
> trait with hundreds of languages around the world--From English to Kalapuya.

Again, I'm not just talking about one feature, but about a whole bundle of
features.  It is this bundle which I consider arbitrary enough to need
historical explanation. I haven't got time to list them all, but I could
mention lexicalisation of aspect, monosyllabicity, lexical tone, restricted
set of syllable-final consonants, classifiers, VSO order, isolating
morphology, etc..
> 

> Contrast this with a real morphological comparison--Indo-European
> verb agreement suffixes:
> 
> 		Greek		Sanskrit	Russian
> 
> 	1st sg.   -mi		 -mi		 -yu
> 	2nd sg.	  -si		 -si		 -eS
> 	3rd sg.	  -ti		 -ti		 -et
> 
> That's sufficiently arbitrary to constitute a strong argument for
> genetic relationship.  I'd guess you would still say of this
>  
> > Then this is still what I would call
> > "lexical", content-based comparison, rather than morphological comparison.
> 
> But if so, you're using the word "morphological" an unusual and kind of
> confusing sense. 

Possibly, but I haven't formally studied in this area for some years.  My
apologies.
> 
> But you do have a good question, if we leave the term "morphological"
> out of it:
> 
> > > >In other words, given that a large group of languages/dialects of
> > > >East/Southeast Asia contains striking mutual phonological, syntactic, and 
> > > >morphological similarities, are these similarities
> > > >
> > > >a) accidental;
> > > >b) borrowed from one another;
> > > >or c) signs of a family relationship ?
> > > >
> > > >Why is there, apparently, such a strong tendency for historical linguists to
> > > >assume that lexical similarites are signs of a genetic relationship, and that
> > > >mere "typological" similarities are something with no genetic relevance ?
> 
> > Right. But, likewise, in the case of Thai and Chinese, there are a whole lot
> > of phonological, syntactic and (I would say) morphological features that are
> > shared.  Why should that not need a historical explanation ?
> 
> It does need historical explanation, but the explanation is borrowing,
> not genetic relationship.  How do we know?
> 
> We know that typological features like this do spread
> across genetic boundaries fairly easily, and often can prove that
> it has happened.  For example, the languages from Japanese down through
> Dravidian, including Altaic, Tibeto-Burman, and Dravidian languages,
> have strikingly similar SOV agglutinative structure.  If you look
> at modern Indic languages, which we know to be Indo-European, you can see
> that over time they fit more and more into this pattern.   Similarly,
> in the Southeast Asian case, we can show (using real morphological
> as well as lexical comparison) that Chinese is related to Tibeto-Burman,
> and Mon-Khmer to Munda (spoken in eastern India).  T-B and Munda
> are SOV agglutinative-type languages, Chinese and Mon-Khmer show
> the typical Southeast Asian SVO isolating pattern.

OK.  Now we're getting down to brass tacks.  Maybe you're right, but just
because this type of borrowing may have happened somewhere else, it doesn't
mean it had to have happened elsewhere.  After all, I seem to remember someone
saying that just because numerals in Indo-European languages didn't get
borrowed, it doesn't mean that they couldn't get borrowed in Tai.
> 
> Your question, if I understand it, is, couldn't Chinese, Tai,
> Mon-Khmer, etc., be related languages that have retained their
> original syntactic structure, but replaced so much of their
> vocabulary that they no longer appear to be related if we only
> look at lexical evidence.  But where would you suppose that the
> new vocabulary would have come from?  It would have to be either
> built out of old native vocabulary, in which case we'd expect to
> be able to see how that happened, or else borrowed--but borrowed
> from where?  But if we assume that the structural patterns were
> borrowed--which, as I've said, is a phenomenon that we see quite
> often--then we don't have any such problem; we can explain both
> the structural similarities and the lexical similarities and
> differences.
> 
Well, you may be right.  But, again, I'm not talking about one single feature,
like SVO order, but a whole bundle of (supposedly) unrelated features.

We also know that lexical items can be borrowed, and I don't think you have
established that this happens less readily than borrowing of phonological,
syntactic, and (do you like this word ?) metamorphological features.

It is at least feasible that ethnic migrations and conquests resulted, in some
cases, in the merging of two languages via Creolisation -- perhaps with the
conquering people retaining formal phonological, syntactic, and
metamorphological features from their original language, and the conquered
people contributing the bulk of the core vocabulary.

> Scott DeLancey			delancey@darkwing.uoregon.edu
> Department of Linguistics
> University of Oregon
> Eugene, OR 97403, USA
>  
Peter Zohrab

