Hypercorrection of /w/-/hw/

[Arnold M. Zwicky]

On May 23, 2004, at 5:43 AM, Dennis R. Preston wrote:

>> arnold,
>
> The lexicalization argument is good, but that it is a carryover from
> casual speech is a bit odd, especially if we associate fast/casual
> speech and, presumably, reduced phenomena it. Exclamatories are not
> reduced but usually strongly pronounced. I'm still a little puzzled
> by the motivation.

casual speech is not the same thing as fast speech or, for that matter,
hypoarticulated speech.  as i use the term, "casual speech" is a label
for a *style*.  "g-droppin'" is a feature of casual speech in english,
and often occurs in notably slow speech (for certain speakers in
certain contexts), and can co-occur with "local" hyperarticulation
(like audible release of word-final t) in fast speech.  for many
speakers who have ultrashort (or entirely missing) unaccented vowels in
the first syllable of words like "Columbus", "police", and "believe",
this feature belongs to casual speech and occurs freely in slow and
emphatic speech. (for many other speakers, like me, this feature *is*
associated with fast and hypoarticulate speech.)

note that in my original posting i said, of hw- versus w-, that the
latter tended to appear in unaccented positions *and* in casual speech.
  maybe that was over-subtle.

so, i'm saying (once again) that things are more complicated than you
might have thought at first.  the linguistic features -- w- for hw-,
g-droppin', fricative release of final t, C'lumbus vowel shortening,
etc. -- are "just stuff", just bits of language, without any intrinsic
connection to (in)formality, rate (but see below), attention to
articulation, or for that matter social variables or aspects of
personal "style" (performance of a persona) in context.  the very same
bit of stuff can do several different kinds of work for a single
speaker, and (as we all know) can do different kinds of work for
different speakers.

it's the "doing several different kinds of work for a single speaker"
part that may be especially hard to appreciate, though examples are all
over the place.  for me, w- for hw- is associated both with fast and
hypoarticulate speech *and* with casual speech (independent of rate and
attention to articulation), and i believe that this is a very common
system.  the (considerable) literature on g-droppin' identifies a large
range of work that this feature can do, and there's no reason to think
that it does only one thing for each person.  finally, research by rob
podesva at stanford shows that fricative release of final t can serve
wildly different purposes, depending on context; for a gay male med
student that podesva studied, in professional settings this feature
(which is a kind of hyperarticulation) seemed to serve as a display of
competence and care, while in social settings it co-occured with a
package of other features (including a *fast* speech rate) that
together seemed to display a flamboyant persona.  (like a great many
english speakers, i have the first use but not the second.  i seem to
display flamboyance mostly lexically; remarkably, i have gotten this
far in my posting without using "extravagant(ly)", though i just used
"remarkab(ly)", another favorite of mine.)

in any case, w- for hw- in exclamatory "why" and in "whoa" seems
entirely reasonable, since neither of these words is likely to occur in
formal speech.

now for the really serious complication.  let's ask where casual speech
and formal speech variants come from.  they don't just drop from the
sky; they must have a history.

well, sigh, the obvious source for casual speech variants is fast and
hypoarticulate speech (for phonological variants, this means lenitions,
serving ease of articulation), and the obvlous source for formal speech
variants is slow and hyperarticulate speech (for phonological variants,
this means fortitions, serving clarity and ease of perception).  the
natural source for w- for hw- and C'lumbus vowel shortening is in fast
and hypoarticulate speech (probably g-droppin' as well, since the velar
gesture requires more displacement from the resting, coronal, position
of the tongue than the alveolar gesture does -- but the history of this
variable is complex, and there might well have been multiple
contributions to its appearance).  the natural source for fricative
release of word-final t is in slow and hyperarticulate speech.

but once these variants appear, they are open for association with
speech styles, independent of their origins in speech rate or attention
to articulation.  they are, of course, also open for association with
all sorts of social and individual properties -- with masculinity or
femininity, with an adolescent identity, with a particular local social
group, with a persona, whatever.  sometimes, these associations become
conventional and shared by members of a group.  that's what i'm
suggested happened with w- for hw-:  though it almost surely started
out in fast, hypoarticulate speech, it became associated with casual
speech (which is often, though not necessarily, fast and
hypoarticulate), just part of the package of informality.  as a result,
casual speech will show significantly higher rates of w- for hw- than
formal speech.  and words that are almost entirely confined to casual
speech are open to lexicalization with the casual variant.

a final note, for the lexicographers here (should any have persevered
to this point):  exclamatory "why" is another one of those cases where
linguistic change has proceeded so as to split one etymon into two
clearly distinct lexical items.  (there are thousands of these.
exclamatory "boy" and "man".  discourse particles like "look" and
"say".  and so on.)  exclamatory "why" (a discourse particle, by the
way) differs from "interrogative "why" not only (grossly) in meaning,
but also, for some speakers ("sincere" hw- speakers) in pronunciation.
actual dictionaries are etymologically arranged (a scheme that doesn't
force lexicographers to decide when uses of a word have diverged
sufficiently to constitute two distinct items), but often at the cost
of lumping together items that are clearly distinct for speakers --
even to the point of putting infinitival "to" in the same entry as the
preposition "to" (i see that both NSOED2 and AHD4, the two dictionaries
i have at home, do this).  this is a principled decision, but to my
mind an awkward one.  it encourages users of the dictionary to think
that "words" are primarily orthographic entities -- infinitival "to" is
the same word as the preposition "to" -- though some "words" are
inexplicably split into separate entities (AHD4 has four entries for a
noun "pen").

i know, i know, this is a very old issue.  i'm just annoyed by some of
the consequences of the usual decision.  i'm now confronted by people
-- highly educated people, attentive to language -- who tell me that "I
don't want to" (and all other occurrences of stranded infinitival "to")
is ungrammatical, because you can't end sentences with a preposition in
english.  it is to weep.

well, i'll get *that* off my chest in another posting, eventually.

arnold (zwicky at csli.stanford.edu)