THEORY: the Obligatory Contour Principle [was Re: Tallefkeul: tones and whatnot]

Quoting Christophe Grandsire <christophe.grandsire@...>:

> En réponse à Marcus Smith <smithma@...>:
>
> > One of the major applications of the OCP was to explain why semitic
> > trilateral roots never have adjacent identical consonants. (Or at
> > least Arabic, I won't swear to the rest of the family.) That is,
> > you could find a root like, say, KTK but never *KKT.
>
> IIRC roots with adjacent identical consonants are pretty well allowed in
> Arabic. See for instance the root WDD, making the verb wadda: to love.
Actually, to my knowledge (and I'm no expert on Arabic morphology),
that's usually considered a case of postlexical spreading.  Although
most Arabic roots are triliteral, some are biliteral.  When these
biliteral roots are mapped onto prosodic templates, the consonants
spread to fill in the remaining templatic spots.  Take a typical
triliteral root, like /drs/ 'study', and compare it to a biliteral
one /md/ 'stretch'.  From /drs/ you can get forms like the following:

  CaCaC    darasa   'he studied'
  CaCCaC   darrasa  'he taught'
  CaaCiC   daaris   'studying'
  maCCaCa  madrasa  'Koranic school'

Now, what would a root like /md/ do with a prosodic template like
CaCCaC?  It has *two* too many consonants, and so it makes do by
spreading:  _maddada_ is the attested form.  This means that even
if we presupposed three consonants, we'd have to assume some kind
of spreading analysis. Okay, fine so far. But in some dialects
of Arabic you get a secret coding game whereby a form like
_darasa_ can come out as _radas_, _rasad_, _dasar_, _sarad_, etc.,
any arrangement of the three consonants to the template besides the
actual one.  But how does a putative biliteral root act? It turns
out that a form like _madad_ can *only* have _damam_ as a
coded form:

  CaCaC     CaCaC
  | |/      | |/
  m d       d m

This would appear to be strong evidence that biliteral roots
do exist, that they undergo spreading, and that the OCP need
not be violated in some apparently 'exceptional' forms.  (Now, this
is only what I recall from a lecture some months ago along with
the always sketchy notes I took in class.  There are many on the
list more expert than I in Semitic studies, so I'd be happy to
hear their thoughts about this.)

 =========================================================================
Thomas Wier
Dept. of Linguistics  "Nihil magis praestandum est quam ne pecorum ritu
University of Chicago sequamur antecedentium gregem, pergentes non qua
1010 E. 59th Street   eundum est, sed qua itur." -- Seneca
Chicago, IL 60637