Re: confession: roots

Oskar Gudlaugsson wrote:
>In our IE languages, and seemingly the Semitic languages too, and probably
>lots of langs all over, the roots are _underlying_. The underlying root
>there would appear to be "phil" and "soph", though I disclaim further
>knowledge of Greek roots.
>...
As to how roots-that-don't-surface evolve, I'd
>guess the original word goes obsolete (diving to the underlying level)
>after a fruitful career of spouting out derivatives. As long as the rules
>of derivation remain stable, the root remains at the underlying level
>through the power of its surface derivations. When the rules of derivations
>change, however, and the lexical structure is reordered en masse, the
>origin of the derivatives may become unclear to the extent that the old
>underlying root is lost completely.....
>fara (vb)    go
>ferð (n)     voyage, trip
>ferðast (vb) travel
>ferill (n)   path of movement; career, life-path
>ferli (n)    process
>far (n)      mark, (im)print ("fingrafar" = finger print)
>far (n)      vessel, vehicle
Minor question:  how many of the apparent suffixes seen above (and in the
others that I've deleted) are still productive?

The nice thing about Indonesian (and probably Chinese, and to some extent
English) is that "root" is equivalent to "word" (Indonesianists prefer to
call them "bases").  A base can occur freely, and has meaning; all sorts of
derivatives are formed with affixes, reduplication, compounding.  Some
bases, especially in very colloquial, or bad, speech can function as noun,
verb, or adjective: _surat_ "a letter"; "to write"; "written". Most bases
are 2-syllable.

Interestingly, there is evidence that somewhere in the past there were many
more monosyllabic "roots".  The only ones reconstructed so far are
onomatopoetic, like *t.uk and *t.ik ''sounds of pounding/tapping"> verbals
*t.ukt.uk 'pound', *bat.ik 'batik-printing'; Ml. titik 'point, dot'.  But
there are a lot of intriguing forms that seem to show a monosyl. root in the
final syllable; sometimes, the initial syllable can be related to known
vocab. or morphology but by no means always.  (These forms are best revealed
in those langs. like Malay, Javanese, Tagalog etc. that retain final
consonants.)  A few exs., all Ml/Indonesian but most have cognates
elsewhere:
a large number of words with /-kang/ that have to do with the legs--
kangkang 'legs wide apart', jengkang 'lift/pull in a leg', jangkang 'walk
with legs wide apart'; pukang 'crotch; thigh' etc..  Quite a few with /-pus/
connoting 'finished':  pupus 'wiped out', hapus 'erase', etc. And many
others. It's harder to find initial syllable "roots", but there is one
impressive set with /pu-/ 'central point, axis?':  pusat 'navel', 'pusar,
putar 'revolve, turn', puntal, pusing 'whirl around'; puntil 'nipple'; and
note pukang 'crotch' above; perhaps in Ml. @mpu 'ancestor; master'? etc.
Various scholars have studied the phenomenon over the years, but aside from
listing the possible roots, not much more can be done, as there is no
evidence that this was a productive process in the proto-language.