----- Original Message -----
Sent: Saturday, April 08, 2000 10:05
Subject: [cybalist] Elaison, r-insertion and the
Glen explains it almost correctly (he's only wrong about
ALL vowels being treated in the same way, see below), but maybe more detail
would help Mark to understand the motivation for R insertion and the mechanism
of its generalisation. It's quite an instructive story.
In non-rhotic accents of English (those in which R drops
word-finally and before consonants) a word-final R was usually retained as a
hiatus filer (a so-called LINKING R) when the next word began with a vowel
(my car is [ka:r Iz] red, but my car's [ka:z] red
and this is my car [ka:]). Because the isolation,
preconsonantal and sentence-final forms of "car" were all R-less, and since only
prevocalic R could be pronounced in non-final positions, the natural
interpretation of the [ka:]~[ka:r] variation was as follows:
[ka:] is the underlying form, and [ka:r] results from
consonant insertion in a special (prevocalic) environment. Note that this is the
reversal of the historical process: the historically older form comes to be
interpreted as derived by segment INSERTION.
A linking R could be pronounced in words which ended in
[a:], [3:] (the central vowel of err, purr,
fur in RP, Bostonian English, etc.), [O:] (as in
more, door, for), [@]
(father, actor), and "centring"
[I@] (near, peer), [U@]
(poor, cure) and [E@] (chair,
bear). The R was also pronounced before any vowel-initial
suffix, as in starry [sta:rI], chairing
[tSE:@rIN], pouring [pO:rIN], dearest
Since phonological processes are normally conditioned by
phonological factors, and not by the (historically based) spelling of the words
affected, the R-insertion rule as stored in the speakers' brains did not take
the form "Pronounce all orthographic Rs if a vowel follows", but
"Insert an [r] after any stem-final non-high long vowel
([a:], [O:], [3:]), schwa ([@]) or a diphthong ending in a schwa ([I@], [U@],
[E@]) IFF the word or suffix that follows begins in a vowel."
(Note: After the high vowels [I], [i:] and [u:] and all
upgliding diphthongs the glides [j] and [w] are used as natural hiatus fillers,
and R is never inserted.)
The problem with that rule is that it overgeneralises
somewhat when compared with its orthographically conditioned version: there are
English words which end in one of the vowels mentioned above, but have no
etymological or orthographic R at the end: pa [pa:],
draw [drO:], sofa [s3Uf@],
idea [aI'dI@], Anna [An@] etc. However, an
automatic phonological process is likely to be more successful in the long run
than a rule which requires the speaker to recall the orthographic form of
the word to decide how it should be pronounced in a given context. So despite
the angry protests of linguistic pedants in "Letters to the Editor", the rule
came to be applied wherever its condition was satisfied, even if a word had no
final R in its spelling:
Joanna-r-is a nice girl.
I had no idea-r-of that.
The Shah-r-of Persia.
I [sO:r Im] at the p[a:]ty.
I saw-r-a few nice draw-r-ings
The best-known city in Nevada-r-is Las
A sofa-r-and a table.
Of course you should not imagine that speakers who use
this INTRUSIVE (non-historical) R always say idea[r] for
idea and law[r] for law The
[r] is inserted ONLY before a vowel, NOT, e.g., in "my idea was different", "it
was my idea!", or "it's against the law").
Hope this helps,
PS. I've decided to start a separate
phonetic/phonological e-group called phoNet. I'll post more details later. --
> I've never quite assimilated
the lectures as to why this happens.
> Glen: generalisation, I'd imagine. The final phoneme drops out of use
> Elaison plus
to break up two side-by-side vowels (just as in French). Then,
generalisation sets in and all side-by-side vowels are seperated with this
> phoneme rather than the usual glottal stops or semivowels.