Eric Christopherson wrote:
> On Dec 5, 2007, at 7:07 PM, T. A. McLeay wrote:
>> In English, the high front rounded vowels were unrounded towards  
>> the end
>> of the Old English period. Mid front rounded vowels were either lost
>> much earlier, or generally not written. Decent (;) dialects of English
>> have since re-created them from things like [u:] and [@:].
> A rounded [@:]? Does that belong to the phoneme which in rhotic  
> dialects is /r=/? In which case, I wonder if that's why German /2/  
> and /9/ sometimes get pronounced in English as /r=/, e.g.  
> _Göthe_ /"gr=t@/, _danke schön_ /"daNk@ Sr=n/.

I don't think it has much to do with the fact that (some) Australians,
Kiwis and probably Londoners have a vowel somewhat like [2:] or [8:] for
rhotic /r=/, but instead the same thing which motivated the rounding
also motivates associating /2:/ with the more dominant unrounded [@:]
variant of /3:/. This gets a bit technical; I don't know how much you
know about phonetics, so I'll try to explain everything. I apologise in
advance if that's too much or too little; it's probably both unless you
know this already. The pretty pictures at the end of this wants to be
viewed in a fixed-width font, so if that's not your default, copy it
into Notepad (or some other plain-text editor).

As I'm sure you know, sounds are made up of waves. Pure tones are simple
sound waves, and can be described accurately with a single frequency (by
definition). More complex sounds, like vowels, use more complex waves
that need to be described with more than one frequency/resonance/
formants. The lowest of  these, the fundamental frequency (denoted F0)
determines pitch, and is used contrastively in tonal languages. Vowel
quality, however, is determined by a few higher ones. In most languages,
these are F1 and F2 (these are called the first and second formant, or
just eff-one, eff-two). Some languages also consider F3. (These formants
are always in order of frequency, with low numbers corresponding to
lower formants, but two adjacent formants can essentially overlap, more

F1 corresponds pretty simply to vowel height: high vowels have a lower
F1, and low vowels have a higher F1.

Vowel backness is represented in F2; back vowels have a low F2, front
vowels have a high F2. Because F2 picks up wherever F1 left off, low
vowels that have a higher F1 will cause F2 to be a bit higher, too, but
this doesn't mean the vowel is less back than a higher back vowel.

Vowel roundedness is represented by both F2 and F3: rounded vowels have
a lower F2 and a lower F3 than the corresponding unrounded vowels. It is
for this reason that most languages don't contrast rounding, but instead
round (non-low) back vowels and keep the rest unrounded. Hence, the
speakers of most languages can pretty much ignore F3 because its
relative value will be implied by F2. But German and French speakers who
make the contrast will obviously need to pay attention to it.

Now then, this becomes relevant to the "Göthe" thing because a German
producing [e] might say something with a mid F1, a high F2, and a high
F3. When they produce [2], they might say something with a mid F1, a mid
F2, and a low F3. Us non-rhotic English speakers proceed to ignore the
low F3 and hear a mid F1 and a mid F2. A vowel like this can correspond
to one of two vowels (x = unrounded, o = rounded):

    If F3 is not especially low:        If F3 is low:
      ---------------------              ---------------------
      \                   |              \                   |
       \                  |               \                  |
        --------x----------                ----o--------------
         \                |                 \                |
          \               |                  \               |
           ----------------                   ----------------
            \             |                    \             |
             \            |                     \            |
              -------------                      -------------

(If these piccies don't come up for you, there's two vowel charts. The
one on the left shows a central unrounded vowel, whereas the one on the
right shows a central-to-front rounded vowel.)

Hence, we non-rhotic speakers are inclined to associate foreign [2:] as
our own /3:/, regardless of whether /3:/ is rounded or not. (Today the
IPA symbol [3:] is defined as being contrastively unrounded, but when
English IPA systems were designed, by the 1980s, [3] was merely an
alternative mid-central vowel contrasting with [@] but otherwise just as
adaptable in quality.)

Now, whether Americans using /r=/ for foreign [2:] is because they're
influenced by similar happenings I'm not entirely sure. I would be
leaning towards the first because I think if you plot F1 against F2 of
/r=/ you end up with something in the vicinity of [U].

(Now, the Australian and Kiwi habit of using [2:] for /3:/ can be
explained by noting that we're not in the habit of paying attention to
F3, so it can be changed quite easily, and noting further that we've
developed quite front rounded vowels for our /u:/ and the offglide of
our /ou/. Speakers are lazy and uninclined to use two methods to get the
same effect (i.e. a mid F2), and so rather than using front vowel +
rounded lips for /u:/ but a mid vowel + unrounded lips for /3:/, we've
gone for the front vowel + rounded lips in both cases.)

> A related thing I've been wondering: How rare is it for front rounded  
> vowels to become back rounded vowels? I don't think I've ever run  
> across that sound change, but it seems plausible to me.

It's happened in Old English -> Middle English before liquids, which
were probably dark. e.g. much <- mycel, worry <- wyrgan.

In Mongolian, vowel harmony operates between two groups called
non-pharyngeal (e u o) and pharyngeal (a U O). These were formally
called front and back. Also /U O/ are spelt with the same cyrillic
letters used for Russian /u o/ (i.e. у о), whereas /u o/ are spelt with
the same letters used for /y 2/ in turkic languages (i.e. ү ө). I don't
know if the old names for these groups and the orthography represent a
historical pronunciation or not.