wheel barrels?

Dennis R. Preston preston at MSU.EDU
Wed Aug 11 16:45:35 UTC 2004


Now let's not be rash here in throwing out the 
phonological evidence in favor of the 
folk-etymological explanation; there's room (I 
would say necessity) for both. I might go so far 
as to claim that the vast majority of folk 
etymologies are rooted in (or at least abetted 
by) phonological matters.

We all know that post-vocalic liquids, even when 
they can be construed as onsets, are troublesome. 
Those who are r-vocalizers often end up with an 
inglide replacement (here íhI-uh) or a 
lengthening of the previous segment (car í caah). 
Those who come from better backgrounds like me 
and are l-vocalizers show similar patterns: 1) an 
upglide replacement (Bill í bIw). Since this is 
an up and back glide, it is predictably round, 
and not the result of the roundness of the 
previous vowel, as Tom might have it; and 2) a 
total loss, perhaps lexicalized (help í hEp).

With these facts in mind, let's look at how a 
perfect speaker of US English like me might be 
misled about the object in question. We start with

hwiyl bEr row

(Those who do not have /hw/ and who do not 
neutralize /ae/ and /E/ before /r/ have my 
sympathies; I said after all that I was 
representing a perfect speaker of US English.)

Stress assignment (taking as many cycles as you like) gives us

1 hwiyl   2 bEr  3 row

Stage 1 of vowel reduction yields

1 hwiyl   2 bEr   3 row

in which the vowel of the weakest stressed 
syllable is still recognizably /ow/ but 
considerable reduced. It crucially, however, 
retains roundness.

Stage 2 of vowel reduction gives us

1 hwiyl   2 bEr   3 ruh

where /uh/= schwa.

Just for completeness (and since we will need it 
later), let's also note that l-vocalization will 
apply to the coda-l in the /hwiyl/ syllable, 
yielding

1 hwiyw   2 bEr   3 ruh

Finally, final (weak) vowel deletion (given the 
appropriate phonetic environment, here following 
an /r/ onset linked to an /r/ coda in the 
preceding also relatively weakly stressed 
syllable) yields

1 hwiyw   2 bEr:

where r: equals a lengthened /r/ (and still gives 
us a third mora, for mora-counters among you).

Of course, the w-glide of the first syllable 
might be deleted in the environment before /b/, 
but that does not affect this discussion.

Also, just so those of you nonstandard speakers 
who also have r-vocalization will not be left 
out, we could add that process before weak vowel 
deletion to get

1 hwiyw 2 bEuh

or perhaps

1 hwiyw   2 bE:

although please note that a process I have not 
described here (coalescing the coda and onset /r/ 
of the final and penultimate syllables) is 
required for these last forms to be derived.

Now let's suppose that we have an underlying 
'wheel barrel,' skipping all the procedures given 
above for the first syllable, since they are the 
same. At the stage of l-vocalization we get

1 hwiyw  2 bEr  3 row

Since there is no preceding vowel in the last 
syllable, the default realization for the 
vocalization is in the /ow/ or /U/ territory. The 
result, of course, it that beginning at that very 
early point in the derivation and from there on 
out the phonetic realizations of "barrow" and 
"barrel" are exactly the same (in Standard US 
English dialects with l-vocalization).

This does not put larry's notion of a lexical, 
folk-etymological process to rest at all. 
'Barrels' are much more common things in 
everyone's experience than "barrows," and r- and 
l-vocalization are common features of child 
language (even in areas where such vocalization 
is not common in adult speech).

Perhaps even more importantly related to larry's 
argument are the notions of hypercorrection (and 
liason, not discussed here) which arise in 
conjunction with vocalization in general.

L-vocalizers are often chastised by the ignorant 
for their perfectly standard behavior and made 
sensitive to their "l-dropping." This makes them 
suspicious that final schwa-like sounds may be 
hiding something that the hoity-toity prefer to 
realize phonetically. A final /uh/ in whatever 
the last segment of the "wheelbarrxx" word might 
be could lead those made linguistically insecure 
to go for a segment. Since /l/ is a segment 
they've been beat up about, why not go for it? If 
this scenario has any value, then "wheelbarrel" 
is no different from hypercorrect "ideal" for 
idea, but it is even stronger at a 
lexical-semantic level since 'barrels' exist and 
'barrows' are little known (just like there are 
'sparrows' and 'grass").

dInIs



>At 10:41 AM -0400 8/11/04, Thomas Paikeday wrote:
>>Bethany,
>>
>>FWIW, here is my explanation of why you hear "wheel barrow" as "wheel
>>barrel":
>>
>>The "l" sound in that position is rounded, so is "w". The "-ow"/"-el"
>>confusion, I believe, is borne out in similar phonetic contexts including
>>"-al", "-il", "-ol", -"ul", and "-yl" if someone will supply examples in
>>support of or against this claim by a non-phonetician.
>>
>>TOM PAIKEDAY
>>www.paikeday.net
>
>Tom, I'm not sure I buy this, however persuasive the phonetic argumentation.
>
>My wife purchased a wheelbarrow earlier this summer and since then
>has referred to it consistently as a wheelbarrel.  (This surprised me
>because she's from Greenwich, CT and doesn't have all that many
>"folk" pronunciations in her dialect.)  I just checked and she
>confirmed that she (like others who have commented) would always
>*spell* it as "wheelbarrow" but usually *pronounces* it as
>wheelbarrel.  She claims (essentially like Tom) that it's "easier to
>say" as 'barrel", but when I grilled her on whether she'd say "bow
>and arrel" because it's easier than saying "bow and arrow", she
>acknowledged she'd be extremely unlikely to do so.  So I think the
>folk etymological link with "barrel" is crucial in the former case,
>whatever the phonetic motivation.
>
>Larry


--
Dennis R. Preston
University Distinguished Professor
Department of Linguistics and Germanic, Slavic,
        Asian and African Languages
Wells Hall A-740
Michigan State University
East Lansing, MI 48824-1027 USA
Office: (517) 353-0740
Fax: (517) 432-2736



More information about the Ads-l mailing list