Q. Were there any prefixes in PIE?
A. The usual answer is negative, but this "No" must be qualified. For
one thing, the distinction between compounds and prefixations (or
suffixations, for that matter) is hardly clearcut. A frequently used
first member of compounds may easily lose its lexical independence and
never or rarely appear on its own. Eventualy it evolves into a true
prefix. For example, the negative particle *ne, often compounded with
PIE adjectives, lost its vowel in composition, changing into *n.-, which
to all intents and purposes behaves like a prefix already in PIE (and a
fortiori in the daughter languages).
Anyone who says that it can't be considered a prefix because
reconstructed PIE has no prefixes is guilty of circular reasoning. Cf.
*dus- 'ill-, mis-, badly', which never occurs in isolation (though it's
probably related to Gk.déomai 'lack, stand in need of' < *déus-e/o-,
Skt. dos.a- 'fault, inconvenience' < *douso-, and a few other words).
Also *(h1)sú- 'well-, good-' is attested as an independent adjective
(and a derived substantive) only in Hitt. assu- 'good (n., adj.)', but
elsewhere in Anatolian we have <wasu-> instead, reflecting *h1wé(:)su-,
which does occur as an independent word in several branches. Perhaps,
then, *h1sú- is an irregular contraction of the free-standing adjective.
Again, if one argues that a prefix derived from an independent lexeme
can't be a "real" prefix, then no prefixes in Latin, Greek, English etc.
are real, since they all go back to prepositions, independent particles,
or even obscured lexical items -- a patently absurd conclusion. If Jens
Rasmussen's "o-infix" theory is correct, some prefixes (whatever their
ultimate origin) must have existed already in pre-PIE.
----------
Q. If derivational suffixes in PIE are basically of the shape *-(e)C- or
*-e/o-, what about *-to-, *-no-, *-ro-, *-tlo-, etc.?
A. These were originally combinations of *-(C)(e)C- with the thematic
vowel *-ó-, yielding *-(C)C-ó-, eventually fused together into a single
morpheme, just as the combination of the thematic vowel *-e- with the
_following_ collective ending *-h2, came to be reanalysed as unitary *-ah2-.
----------
Q. What is the function of the thematic vowel *-e/o- in nominal stems?
A. It seems to have formed adjectives of origin, belonging, or a similar
kind of abstract connection (*X-o- = 'having to do with X'). For
example, RV avya- means 'coming from sheep (<avi->)', and Lat. septimus
means 'seventh' ('connected with the number seven (<septem>). Hence the
hypothesis (discussed here a few months ago) that, e.g. the verbal
adjectives in *-tó-/*-nó- go back to *-(e)nt-é/ó-, e.g. *kWr.tó- 'done'
('coming from the doer, *kWr-ent-'). Adjectives can easily undergo
substantivisation, so that we also have a great number of thematic
nouns, often with contrastive accent distinguishing them from related
end-stressed adjectives. Some nouns and adjectives with final *-o- are
etymologically opaque, which means that their hypothetical base is
unattested on its own. For example, *wl.kWo- 'wolf' seems to contain the
suffix *-o-, but we can't identify the underlying root (*welkW-?).
----------
Q. When is the thematic vowel replaced by *-i/j-?
A.
(1) In composition, cf. Lith. mai~nas 'exchange' < *moino-s vs. Lat.
commu:nis 'common, general' < *kom-moini- 'shared mutually'.
(2) When a thematic adjective is derived from an already thematic base,
e.g. RV as'vya- 'pertaining to horses' from <as'va-> horse.
(3) Before certain suffixes, e.g. diminutive *-ko- (yielding *-iko-).
Note also the alternation of *-e-h2 with *-i-h2 in feminines.
----------
This is just for starters, as a means of provoking further discussion.
Piotr