Jim Henry skrev:
> On Fri, Jul 10, 2009 at 8:38 AM, Brett Williams<[log in to unmask]> wrote:
>> syllables, and a fair number of allowed initial clusters.  I don't
>> have any idea how to go about picking which clusters to allow, though,
>> except to pick ones that seem sayable to me-- which in my ignorance
>> would probably just mostly be ones I'm already familiar with from the
>> languages I know.  Are there any general principles I can follow?
> See this thread on "Most common consonant clusters" from last August:

I have a hunch that here is an implicational
hierarchy something like the following
with combination types higher in the list
usually being allowed before those lower
in the list.

  1. YC
  2. CY
  3. YCY
  4. NC
  5. LC
  6. CL
  7. NCY
  8. LCY
  9. YCL
10. NCL
11. LCL
12. SC
13.-24. Like 1.-11. but SC instead of C
25. CT
26.-48. like 1.-12., 19.-30. but CT instead of C
49. CC


* Y = semivowel
* C = obstruent
* N = nasal, usually homorganic with following C
* L = liquid, /r/ more often than /l/
* S = sibilant
* T = coronal obstruent (dental/alveolar)

Of course various specific cooccurrence restrictions,
like agreement in voicing between adjacent
obstruents may apply.  It is also pretty
common that geminates occur even if clusters of
sounds of the same type but with differences
in articulation aren't.

I'm not sure where YY LY YL YLY NY YN YNY
go relative to 1.-6.  My hunch is that they
may go before them or parallel with them depending
language.  Note that NN LN LL NL may be quite
restrigted (more likely so going from left to right
in the order listed: they are often subject to
assimilation into geminates (with L winning out over
N) or simplification. LL and NL are often subject
to insertion of a (voiced) stop homorganic with
the first sound in the pair.

NB I've got no proof for this: it's just a hunch
acquired from studying various languages and
linguistics through 30 years.

One language I know which partly breaks
the tendency is Finnish: it allows 2.-5.
SC and 14.-17. plus /ts/ and /tk/ and
geminates, dubtfully NCY and LCY but
no others. (Counting /v\/ as Y of course!)

/BP 8^)>
Benct Philip Jonsson -- melroch atte melroch dotte se
  "C'est en vain que nos Josués littéraires crient
  à la langue de s'arrêter; les langues ni le soleil
  ne s'arrêtent plus. Le jour où elles se *fixent*,
  c'est qu'elles meurent."           (Victor Hugo)