Q: "cognate" by borrowing

[John Hewson]

----------------------------Original message----------------------------
On Mon, 8 Jun 1998, Jim Rader wrote:
 
> ----------------------------Original message----------------------------
> I recall that Eric Hamp avoided the use of the word <cognate> as
> a linguistics term, though I was never sure if he did so because he
> had a very narrow definition of it, or because he felt its ambiguity
> rendered it useless in precise characterization of  linguistic
> relatedness.  My own sense is that <cognate> ought to refer only to
> strict linguistic sibling relationships and to forms that are exactly
> matchable segment for segment; otherwise it's not distinct from
> <related>.  But that's only a personal predisposition.
 
The question is further complicated by the fact that borrowings, if they
take the form of calques, may also be genuine cognates. This may result in
the reconstruction of words for modern European artefacts in Amerindian or
African protolanguages that long predate European contact. Such calques
can be the result of at least two different processes, as follows.
 
1. If speakers of closely related language B borrow a word from language A
using their own phonology, they may may quite accidentally use
correspondences that are strictly historical. Cree, Fox, and Ojibway have
words for `gun, shoot with a gun' that are perfectly cognate, but if the
order of borrowing were from Fox to Ojibway to Cree, the correspondences
could be phonological accidents, given, for example, that Plains Cree
does not have the `sh' of F and O, and would automatically replace it with
`s'. Proto-Algonkian *_pa:shkesikani_ `gun'. F pa:shkesikani, O
pa:shkisikan, C. pa:skisikan.
 
2. If the transmission were different, the phonological accident is no
longer probable (O has both `shk' and `sk'), and what has happened
is that the borrowers have recognized the word formatives of the source
language and replaced them with the cognate formatives of their own:
_pa:shk_ `burst'; _esi_ `by heat'; _kani_ `nominal'.  This is clearly
what has happened with the Cree, Menomini, and Ojibway words for `church',
which are perfectly cognate, but composed of the easily recognized
formatives `prayer + building'. (Strictly speaking it is the formatives,
not the words, that are cognate).
 
English borrowed many nautical words from Dutch which were cognate, but
without calquing, so that we now have both `boom' and native `beam'. If
calquing had taken place we would have had `shipper' instead of `skipper'.
Such `uncalqued' borrowings from closely related languages are a different
kind of problem that is especially notable in the Australian data, as
pointed out by Harold Koch.
 
 
 
John Hewson                                     tel: (709)737-8131
University Research Professor                   fax: (709)737-4000
Memorial University of Newfoundland
St. John's NF, CANADA A1B 3X9