Hypercorrection of /w/-/hw/

[Arnold M. Zwicky]

On May 24, 2004, at 3:06 AM, Douglas Bigham wrote:

> Interesting side bit... I never thought I came from a sincere /hw/
> land, but
> after spending this past year *real* collected data, it turns out I do.
> Mostly (that is, for most of my speakers) it's just preserved to
> distinguish "when" from "win" (and less often "where" from "were"),...

let me rephrase this: most of your speakers have hw- (presumably,
variably) only in "when" and "where".  these are two high-frequency
items that fairly often have some sentence accent, so it's not
surprising that they  have maintained a lexical hw- as other items with
historical hw- have been reanalyzed as having lexical w-.

your explanation in terms of contrast maintenance strikes me as very
dubious, however.  many historical hw- words have minimal-pair w-
counterparts ("which"/"witch", "whales"/"Wales", "why"/"Y") but don't
have the hw- maintained; many other neutralizations have proceeded
merrily on their way even when they result in significant ambiguities
("pen"/"pin", "Dawn"/"Don"); and anyway in the case of "when" and
"where" the chances for ambiguity in context are vanishingly small.

> ... but (and here's the fun part), I've also heard it pop up a couple
> of times in "well", in all age groups. Go figure.

this is actually not so surprising.  suppose you're a speaker for whom
the neutralization is complete, or nearly so.  for you, the hw-/w-
distinction serves little or no lexical function.  but you're perfectly
capable of detecting the phonetic distinction in the speech of others
around you (just as you are able to detect fricative release of
word-final t, which is utterly sub-phonemic).  if you detect it, you
can assume that it's just random noise, or you can hypothesize that
it's doing something non-lexical -- marking something of discourse,
social, etc. relevance.  most of the occurrences of hw- you hear are
going to be in accented words, and these will be especially noticeable
in words with strong emphatic accents.  so it's a reasonable hypothesis
that hw- is marking emphasis.  in which case, high-frequency w- words
that often occur with emphatic accents, like the discourse marker
"well" (but not, i suspect, the ordinary adverb "well"), can have their
emphatic character reinforced by hw-.

(it would be perfectly possible for some speakers to end up with *both*
lexical hw-, in some words, *and* paralinguistic hw-.)

of course, once some speakers have made this reanalysis, the new system
can spread on its own.

once again, there's an important lesson here: you can record
occurrences of some feature but still not know what that feature is
doing for the people who use it.  the feature is "just stuff".

by the way, i'm pretty sure i've heard emphatic hw- in "what"
(especially in the one-word incredulity question "What?!") from
speakers who otherwise  have no hw-/w- distinction.  (i don't have
notes, however.  but it would be worth looking into.)  this would be
just like emphatic hw- in "well", though with a different history
behind it.  (the speakers are, of course, ignorant of the history.)  i
would not be entirely surprised to come across similar uses of hw- in
the discourse marker "whoa" ("Whoa, dude!").

arnold (zwicky at csli.stanford.edu)