Reversal of merger

[Larry Trask]

----------------------------Original message----------------------------
It is a commonplace of historical linguistics that phonemic mergers
cannot be reversed, a position recently dubbed "Garde's Principle" by
Labov (1994) -- somewhat unfortunately, since "Garde's Principle" was
already in use for something completely different.
 
Nevertheless, a number of apparent counterexamples have been reported
in the literature, and proposed explanations for these are not in
short supply.  So far, I have tracked down five putative explanations
for apparent reversals of merger, as follows:
 
(1) The merger occurred only at the phonetic level, and speakers
retained different underlying representations in their heads, which
they could later provide once again with differing phonetic
realizations (Halle 1962).
 
(2) The merger took place, but just one of the two merged segments
had a distinctive phonological role in the language, and speakers
were later able to separate out the two cases and de-merge them
(Michelena 1957, 1961).
 
(3) The merger took place in the prestige variety, but not in another
variety of lower prestige, and a switch in prestige shows up in the
record as a reversal, since only prestige varieties tend to be well
recorded (Weinreich et al. 1968).
 
(4) The merger occurred only variably, and speakers retained both
merged and unmerged pronunciations, but tended to report only the
merged pronunciation (Milroy 1992).
 
(5) The merger never occurred; instead, there was only a near-merger,
resulting in the usual failure of speakers to observe the objectively
real contrast (Labov 1994).
 
Now my query is this: does anybody know of any additional attempts at
explaining apparent cases of reversal of merger?
 
Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
England
 
larryt at cogs.susx.ac.uk