Synonymy avoidance

[Laurence Horn]

At 10:28 AM -0800 3/8/05, Ed Keer wrote:
>The discussion of 'dope' reminded me of this issue. I
>have a friend who is absolutely convinced that there
>is no synonymy in English (I assume he feels this way
>about other languages too). He looks for meaning
>differences everywhere becuase he wants to be sure he
>says exactly what he means.
>
>For example, he won't use 'since', except temporally
>because he really means 'because'. He's also driven
>himself a little batty looking for the meaning
>difference between 'that' and 'which'.
>
>I've heard some linguists believe there is no synonymy
>in langusge, but I have a hard time understanding what
>that means. Can anybody enlighten me on the issue?
>
Like the homonymy issue, it depends on how narrowly you cast your
net.  The claim (which I associate with Michel Bréal and Dwight
Bolinger, but I'm sure there are there others over the decades who
have taken similar positions) is that there is no true/complete
synonymy, if this is defined not just to take into account sameness
of truth-conditional or literal meaning but also identity in
connotation, register, "quaintness", grammatical properties, etc.,
with the result that the two items, the putative synonyms, would be
mutually substitutable for each other in all contexts without
affecting meaning or "tone" in any way.  So the example we were
discussing a couple of weeks ago, "fridge" vs. "ice box" vs.
"refrigerator", wouldn't be a counterexample, given their difference
in register, even though they all have the same referential meaning
and arguably the same sense.  The idea is that each word must earn
its keep in the (mental) lexicon or it will disappear; Bréal (in the
late 1890s) posited a "law of differentiation" that dictated that the
meaning or use conditions on one of the synonyms (resulting from the
adoption of a loanword, for example) will shift, so synonymy will no
longer obtain; he gives examples like "animal" vs. "beast" vs. "deer"
(orig. = 'animal').   The "Avoid Synonymy" principle, so named by
Paul Kiparsky in the early 1980s, also applies to the meanings of
affixed terms; the result is only one word per meaning slot.  This
can be (and has been) seen as a reflex of a more general "Elsewhere
Principle" that Kiparsky traces back to the great Sanskrit grammarian
Panini.

larry