"Pan-Americanisms"

[Miguel Carrasquer Vidal]

----------------------------Original message----------------------------
I was recently unable to resist the temptation to buy Lyle Campbell's
"American Indian Languages.  The Historical Linguistics of Native
America" (Oxford University Press, 1997), despite the fact that my
budget strictly speaking did not allow for it.
 
I do not regret it.  The book contains very thorough and extensive
information on the history of American Indian linguistics, on the
classification of the languages of North, Middle and South America,
on pidgins and trade languages, linguistic areas, etc., complete with
maps and indices.
 
The polemic part are obviously chapters 7 & 8 ("Distant Genetic
Relationships: The Methods" and "Distant Genetic Relationships: The
Proposals").  Chapter 7 is an first rate resume of the general issues
in dealing with proposals of genetic relationship: lexical versus
morphological comparisons, sound correspondences, borrowings, chance
resemblances, non-resemblances, semantic shifts, nursery/expressive
words, errors of method and errors in the data, etc.
 
Two of the points discussed, however, can be said to be specific to
the field of Americanistic historical linguistics: the so-called
"pronoun argument" (focusing on Greenberg's n- "I", m- "you") and the
"so-called Pan-Americanisms".  It is this latter concept that I have
some specific questions (doubts) about, which I would like to put
before the author and the other list-members.
 
The methodological discussion of these "pan-Americanisms" (pp.
257-259) clears away some of the misunderstandings that the term
might suggest.  It was apparently introduced in Campbell and Kaufman
1980 ("On Mesoamerican linguistics", AA 82), referring to:
   ""widespread forms (so-called pan-Americanisms)" [..] which are
   not (necessarily) genetically related forms but may be due to such
   factors as onomatopoeia, sound symbolism, borrowing, nursery
   formations, universals, and accident".
 
In the same article, Campbell and Kaufman recommend eliminating
"pan-Americanisms" from proposals to link specific groups of American
Indian languages together.  One cannot use one of these wide-spread
forms to prove that language A is to be grouped with language B, if
the same form also occurs in languages X, Y and Z (historical case in
point: Sapir used n- "I" , m- you" to support his Algonquian-Ritwan
proposal, but since these pronouns are wide-spread outside Algonquian
and Ritwan, Michelson's criticism was justified.).  In sum:
   "Two points about so-called pan-Americanisms should be emphasized.
   First, these widespread forms are not necessarily traceable to
   inheritance from a common ancestor; second, these widespread forms
   are not valid indicators of narrower proposed genetic groupings if
   the forms are prevalent in languages not included in the
   comparison."
 
Still, something fishy remains about these forms [why not simply
classify them as onomatopoeic, borrowed, etc., and how can
"universals" be restricted to the Americas in order to produce
"pan-Americanisms"?], and indeed Campbell continues:
   "Finally, the possibility must be entertained that some of these
   widespread forms may actually reflect wider historical connections
   than are recognized at present".
 
I agree with the three points.
 
However, as we move into chapter 8, discussing specific proposals
about distant genetic relationships, I get the feeling that there is
a logical flaw in "pan-Americanism" as grounds for dismissal of
specific forms.  Let's take an example at random, Mary Haas'
Yana-Karuk (Hokan) comparisons.  Campbell says:
   ".. of the 92 look-alikes compared, 13 are onomatopoeic [..], 26
   are short forms [..], 10 reflect semantic latitude [..], 23 are
   widespread or pan-American forms, 15 have little phonetic
   similarity; [2] are suggestive of diffusion [..]; and [1] is a
   nursery form".
 
Unfortunately, the total number of rejects is 90, so I cannot work
out if the categories are meant to be mutually exclusive or that they
overlap.
 
I should say that I haven't seen Haas' data (and that I wouldn't be
able to recognize the "pan-Americanisms" even if I had).  But here we
have a group of 23 sound-alikes being `dismissed' as "pan-
Americanisms" (apparently not of the kind that is transparently
attributable to onomatopoeia or borrowing).  Certainly, the
"pan-Americanism" argument, as explained above, can be used against
each and every one of these 23 forms individually, but can it be used
against the 23 as a whole?  That is my worry.   English "two" and
German "zwei" cannot individually be used as an argument for a close
relationship between these two languages, because we are clearly
dealing with "pan-Indo-Europeanisms".  But what about a block of 23
"pan-Indo-Europeanisms" found as a group in Germanic and Germanic
only?
 
 
=======================
Miguel Carrasquer Vidal
mcv at wxs.nl
Amsterdam