I wrote:
> It occurs to me that in a language
>> with contrastive initial /?/::/0/, it might be the onset of
>> /0/-initial
>> words??  Since IIRC you're familiar with both Arabic and Hawaiian, am
>> I
> more
>> or less right?

From the various replies, it sounds like I was wrong.

<Is there any friction, or is it smooth?>

Again, evidently there is friction. So much for that idea....

I was aware of the Tongan situation, cited by Philip Newton-- there, /?/
goes all they way back to Proto-AN *q (it may well have been *[?] by the
time Proto-Oceanic developed); I'm pretty sure it's lost in all other
Polynesian langs., but some (like Hawaiian) have developed /?/ < *k.

The only cognates I recognize in Philip's list:
To. 'au 'current' vs. au 'me' =
Malay harus *qaRus vs. aku *aku
Haw. IIRC au  vs. a'u

Digressing a bit, there are Austronesian languages (and surely others) that
have automatic [?] onset to vowels, but when you add a CV-prefix, some words
show /?/, while others have smooth transition-- e.g. of two homophones with
initial [?]i... one may go to **ma?i... the other > **mai....  One is always
suspicious of [?] between like vowels.