manaster at umich.edu
Thu Jan 28 20:22:54 UTC 1999
I see we have a failure to communicate. I was assuming that
the question was whether to demonstrate language relatedness
we need to have first established definite regular sound
correspondences (as Trask seems to believe) or not (as I believe).
Rich and Johanna appear to be dealing with a different
question, namely, whether to establish relatedness
we need to look at lexical or morphological items.
This is an entirely separate question from the one I
was addressing. It is indeed the case that some of the
lg families I mentioned were established largely on the
basis of morphological comparisons, but that was not
My point is that, when comparing morphemes and morpheme
sequences (whether lexical or grammatical) it is posible
to (a) insist on regular sound correspondences or
(b) look at phonetic similarities (as I will discuss another
there are other possibilities besides), and I maintain
that many generally recognized lg families
were established using approaches of the latter type--
or some mixture of approaches but not pure (a).
Rich and Johanna also say something about Romance and
Slavic and Semitic being "obvious" even to nonlinguists,
but this misses the point that many connections which
nonlinguists take as "obvious" have turned out to be wrong.
I know many people who think English is related to French
in virtue of words like chef, place, etc., but they are
wrong. It is thus still a significant question for lx
how one decides about the correctness of even the obvious
Back to the real issue. Another example just came to mind: Jeff Leer
Tlingit is related to Athapaskan (as I think everybody
now agrees) but there is a lack of regular correspondences,
which he tries to explain as a result of Tlingit itself
being a mix of several quite different (Pre-)Tlingit
PS. The question of whether there are lg families
which were established on the basis of purely
lexical comparisons is also an interesting one
but I will postpone discussing that one too.
On Wed, 27 Jan 1999, Richard M. Alderson III wrote (and Johanna
Nichols promptly endorsed his statement):
> ----------------------------Original message----------------------------
To challenge Larry Trask's assertion that linguistic relationships are not
> *demonstrated* by phonetic resemblances, although they may suggest hypotheses,
> Alexis Manaster-Ramer lists the following:
> >Examples: Slavic, Romance, Germanic, Indo-European, Shoshonean (later
> >under Uto-Aztecan), Semitic, Malayo-Polynesian, Samoyedic, Finno-Ugric,
> >Turkic, Mongolic, Tungusic
> Let's examine these for a moment.
> For sentimental reasons, I'll start with Indo-European. Sir Wm. Jones may
> noted a "phonetic resemblance" among Sanskrit, Greek, and Latin which piqued
> his curiosity, but we all know upon what evidence he drew his conclusions: The
> "affinity, both in the roots of verbs and in the forms of grammar" of the
> Romance was simply "known" to be a family, given the historical knowledge that
> all these languages were somehow "corrupt" forms of Latin--not merely in the
> phonology but in the morphology. Similarly, tradition held that the Hebrew
> Arabic languages were spoken by descendants of the two sons of Abraham, so the
> similarities in morphology and phonology were simply accepted.
> I believe--I admit I have not read the 1752 monograph--that it was morphology
> rather than phonology that convinced a Hungarian diplomat that the Finnish
> language was related to his own, though he likely noticed the phenomenon of
> vowel harmony first.
> I sincerely doubt that any of the accepted families in the list of examples
> accepted because of similarities in the non-grammatical lexicon alone, or were
> posited only because of such similarities. While early writings may not share
> the rigour of our current discipline, it was clearly "grammar"--that is,
> morphology both inflectional and derivational--that was the deciding factor in
> the success of these hypotheses of relationship.
> Rich Alderson
More information about the Histling