IPA transcription of the American English "bunched" /r/

Question

There are 2 common articulations of /r/ and /r̩/ in American English, one retroflex, and the other dorsal. This phone is called the molar or bunched r. It can be described roughly as a back-palatal or pre-velar approximant that's somewhat bunched up along the left-right axis, and it's often labiodentalized when syllable-initial. RainDoctor cites John Laver as describing it as a "voiced labial pre-velar approximant with tongue-tip retraction".

I've been transcribing it ad-hoc as ⟨ɰᶹ˞⟩ or ⟨ɨᶹ˞⟩, using ⟨˞⟩ as an attempt to indicate the bunching that imparts the characteristic sound.

But this seems like an abuse of notation, and it doesn't specify what the articulatory change is that makes it rhotic; is there a standard way to represent this phone?

(Asking on Linguistics since it's not about the sound per se, but about the notation for transcribing it.)

Answer 1

This "bunched or molar R" sound is widely acknowledged as existing, but has engendered a lot of discussion, not just about its symbol but its exact articulatory nature and even its acoustic distinguishability from a retroflex r.

As you note, this sound is discussed at some length in the original Laver 1994 book Principles of Phonetics. Laver himself did not think a satisfactory symbol existed and proposed the symbol ψ (Trask 1996 gives this as Ψ). Laver also suggested potentially using the tongue-root-retraction symbol ̙ (Unicode Combining Right Tack Below) but in a superscript position -- for which there is no Unicode symbol -- in combination with "an existing approximant symbols of suitable lingual and labial attributes". Tellingly, in my opinion, he didn't give an example of what that proposal would look like.

Martin Ball in a 2011 poster [link to original Word document] has proposed ɹ̈ (alveolar approximant with the centralization diacritic) which seems reasonable, and has the advantages of being expressible in Unicode and not requiring a non-IPA symbol.