Sum: `cognate'

[Larry Trask]

----------------------------Original message----------------------------
In response to my query about the use of `cognate', I received the
modest total of eleven responses.  And (I'm getting used to this)
there was no consensus.  Four people endorsed the strict use of
`cognate', the sense in which no borrowing is involved.  One more
appeared to do the same, but was less explicit.  One person preferred
an even narrower sense of the term, by which cognates are only
cognate if all their segments are cognate.  But two people endorsed
the broad use of the term, by which words which are related by
borrowing may also be described as `cognate'.  The remaining
respondents took no view.
 
So I guess I must write an essay describing the multiple uses of the
term, with a cautionary note about the widespread (?) preference for
the narrower use.
 
I might mention that I already have an entry for `oblique cognates'.
Oblique cognates are words that share an ultimate common origin of
some kind but are not derived from identical etyma.  For example,
English `feather', according to Watkins, derives from PIE *<pet-ra:->,
while Greek <pteron> `wing, feather' derives from PIE *<pt-ero->, two
different formations based upon the PIE root *<pet-> `fly'.  And I
suppose the IE `tooth' words might constitute another such case, since
Latin requires PIE *<dent->, Greek and most of Germanic require PIE
*<dont->, and Gothic requires PIE *<dnt->.  This example is cited
under the entry for `tooth problem', the name of the problem which
these forms illustrate.
 
A couple of respondents drew attention to a variety of complex
scenarios, most of which have no recognized names.  There does exist,
though, the term `loan nativization', applied to cases of borrowing
from closely related languages, in which either the foreign segments
are systematically replaced by the corresponding native segments
(corresponding, that is, according to the regular systematic
correspondences between the two languages), thereby producing
something that looks wholly native, or the foreign morphemes are
systematically replaced by cognate native morphemes, thereby producing
something that looks for all the world like a purely native formation
(this is a kind of calquing).  Some of the examples cited would appear
to fall under this heading, but not all.
 
On `cognate languages', there was general agreement that this term is
no longer in regular use.  However, three people objected to our
standard term, `genetically related languages', on the ground that it
puzzles or annoys biologists and other non-linguists, and they
suggested alternative terms.  Be that as it may, `genetic
relationship' is our established term, so it has to go in, though I
can perhaps add a note pointing out the possibly misleading nature of
the term.
 
A couple of people also raised other issues which I won't pursue here,
since I have several piles of exam scripts waiting for me, and more to
come.
 
My thanks to Sally Thomason, Alexis Manaster Ramer, Claire Bowen,
Richard Coates, Roger Wright, Jim Rader, Harold Koch, Steven
Schaufele, Max Wheeler, John Hewson and Henry Hoenigswald.
 
Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
England
 
larryt at cogs.susx.ac.uk