Re: laterals (was: Pharingials, /l/ vs. /r/ in Southeast Asia)

Trebor:
> And wrote:
> "lj, though, is a different story. As far as I can tell, it is just
> disappearing over time, so that hardly anybody nowadays would say
> /slju:/ for 'slew' (the noun, not the verb), while no young person
> is likely to say 'lure' with a /lj/ (ergo I myself am no longer Young).
> That's word-initial /lj/; I haven't tried asking people how they say
> 'curlew'."
>
> Don't some British people say [&ljumInI@m] for 'aluminum'? And what about
> 'million', /mIlj@n/?
Yes. Word-internal lj is going strong in many parts of Britain.
When I wrote the quoted passage I was thinking aloud a bit too
much, and left out certain segments of the thought.

In the case of _million_, it alternates with a trisyllabic /mIli@n/,
so it could be possible for someone to argue than in [mIlj@n] the
[lj] doesn't represent two segments occupying the same onset.

In the case of both _million_ and _aluminium_, the vowel before the
l is short, so the l could be analysed as sharing the rhyme with
the preceding vowel, leaving the /j/ free to occupy an onset on
its own. L-vocalizing accents confirm that: [mIoj@n], [aoj@mInk@m].
Even _curlew_, with its long vowel, is susceptible to having the
/l/ absorbed into the coda, though. _Fail_ is [f&o]; _failure_
is, for some, not [fEjlj@] but [f&oj@]. So _curlew_ might come
out as [k@ojy] or similiar. In the accents in question, Ls darken
or vocalize only when not in an onset; since their tendency is
increasingly to vocalize before /j/, I conclude that they don't
share an onset with /j/.

--And.