Q: "cognate" by borrowing

[Larry Trask]

----------------------------Original message----------------------------
I have encountered a certain slipperiness in the use of the term
`cognate' in connection with borrowing.
 
Take an example.  Late Latin *<caveola> ~ *<gaveola> `small enclosed
place' is the direct ancestor of Old French <jaiole> `jail', of Old
Spanish <javola> (modern <jaula>) `cell', and of Occitan <cayola>
`cage'.  These Romance words are cognate in any definition.
 
But English `jail' is borrowed from the Old French word, Basque
<txabola> `hut, cabin' is borrowed from the Old Spanish word, and
Basque <kaiola> `cage' is borrowed from the Occitan word.  So: would
you be happy to say that English `jail' is
 
(a) "cognate" with its Old French source?
(b) "cognate" with the other Romance words?
(c) "cognate" with the Basque words?
 
Where your answer is "no", what brief label would you attach to the
historical connection between these words?
 
Existing textbooks of HL are generally rather hazy on this point.  At
least three of them don't even have an entry for `cognate' in the
index -- which I find amazing.  Two or three expressly define
`cognates' as `words (or other elements) having a common ancestry', or
words to that effect, which would appear undeniably to include the
cases of cognation by borrowing, but then go on to imply strongly that
cognates necessarily form part of a genetic relationship.
 
Oh, and one final query.  General dictionaries of English consistently
recognize the phrase `cognate languages'.  Would everybody agree that
this usage is now defunct in favor of `(genetically) related
languages'?
 
Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK
 
larryt at cogs.susx.ac.uk