iron: /@r/ vs syllabic /r/

[Aaron E. Drews]

At 10:50 AM +0800 11/8/00, Laurence Horn wrote:
>As someone weaned on SPE, I'm wondering how treating RP as
>underlyingly non-rhotic would allow you to handle the standard
>alternations and sandhi phenomena (e.g. when a word with a final -r
>is followed by a vowel-initial word, or suffix).


You have to take a theoretical step back to the structuralists, but
incorporate autosegmental notions.  What Giegerich (my supervisor)
has proposed for RP is that /r/ becomes [r] in onset and [@] in
rhyme, or [3] in stressed rhymes.... allophones of the same phoneme
kind of thing.  There is phonetic basis as well as complementarity
for this assessment.

Intervocalically, /r/ is in both onset and rhyme.  By being in onset,
[r] is phonologically obligatory (socially, it isn't, depending on
the orthography/history). By being simultaneously in the rhyme, any
changes like centring diphthongs can be accounted for.  The most
important thing I've discovered is to stop calling it 'intervocallic'
and call it 'ambisyllabic'... (same with American flapping).

This accounts for word-final, pre-vocalic r.  It also accounts for
a-historic r.

At 11:07 AM -0500 11/8/00, Herb Stahlke wrote:
>Is RP like NE dialects in shifting from r-deletion to r-epenthesis?  I
>don't recall whether an RP speaker would say [kjub@ Iz] or [kjub at r Iz].

Neither.  In fact, I doubt there ever was actual deletion (and
therefore no shift to epenthesis).  Instead, [r] (official IPA
alveolar trill)(archaic Scots)  became an alveolar retroflex
approximant (modern Irish English), then the retroflex loosened up a
bit (American), then loosened up a bit more (non-rhotic England).
Somewhere in that progression, /r/ went from a consonant to a glide,
and a glide becomes a vowel in coda, and being a vowel, all sorts of
phonotactics get involved.

In fact, I believe, that pretty much the same thing happened in
proto-American English.  That is /r/ became a glide and is now
realised as a vowel in codas (borrowed from Khan 1976). We just
happened to keep our tongues bunched up a little more, preventing a
merger of /r/ and /@/.

What I'm saying is that the difference between non-rhoticity and
rhoticity (as it is realised in American English) isn't all that
great. In both groups /r/ is a consonant in onset but a vowel-like
segment in rhymes.  The difference is  the actual articulation of the
rhyme segment. This is where traditional definitions of rhotic and
non-rhotic begin to get a bit fuzzy, and me calling RP underlyingly
non-rhotic was based on the traditional definition.

And yes, it is [kjub at r Iz].



--
________________________________________________________________________
Aaron E. Drews                               The University of Edinburgh
http://www.ling.ed.ac.uk/~aaron      Departments of English Language and
aaron at ling.ed.ac.uk                    Theoretical & Applied Linguistics

Bide lang and fa fair  \\  //
                         \\// /
                             /