On 2006-09-12 06:09, pielewe wrote:
>> Why not a neuter ending, given the fact that the formation is
>> hypocoristic?
>
> Two reasons: because the semantics are definitely masculine (heroes
> routinely have names in -o, the word "hypocoristic" conveniently
> papers over this point)
For a variety of reasons, heroes and rulers are often known under what
are etymologically pet names (suffice it to mention Robin Hood and
Attila), and examples of nouns with 100% masculine semantics and neuter
etymology are not difficult to find (cf. Pol. ksia,z.e,). There's no
need to believe that the use of an originally hypocoristic form implies
any disrespect to the hero or casts doubt on his machismo.
> and because there is the problem of the
> indeclinability. Both points are unproblematic on the basis of Nsg -o
> and remain unexplained otherwise.
I'd say that the neuter theory fares even better in this respect. We get
nom.sg. = acc.sg. = voc.sg. -o for free, so there's more _initial_
indeclinability in the pattern and consequently less to explain. Once
the grammatical interpretation of the diminutives had switched to fully
masculine, those atypical -o forms gave the impression that the
formation was simply indeclinable. What's problematic here?
> It is my impression (at least: it *was* my impression 15 years ago
> when I was working on it) that this is typical of the situation: o#
> yields straightforward solutions whereas U# spawns the need to devise
> ad hoc solutions.
>
> Take the 1pl endings. If I remember correctly we have the endings:
> mU#, mo# and me#. These endings look as if they cannot be reduced to
> each other, they all exist. Why assume that mU# and not mo# reflects
> *mos#?
It isn't fair to say that *-o < *-os yields a straightforward solution.
It leaves *-mU unaccounted for (Kortland's *-mom is as ad hoc as
anything). If, however, we accept -mU < primary *-mos (parallel to Lat.
-mus), -me and -mo can be viewed as traces of s-less secondary endings
(it's also possible that -me < *-mes).
> [I do seem to recall though that the historical phonology of Old
> English, Old Norse and Old High German brings you face-to-face with
> pretty complex issues sooner than does the historical phonology of
> OCS or most Slavic daughter languages. After that, Slavic seemed
> almost embarrassingly regular.]
Just one example: Germanic became the very epitome of the embarrassingly
regular after the discovery of Verner's law. Now "everybody knows" that
Grimm's Law came first and Verner's Law came after it (I mean the actual
sound changes, not their linguistic formulations) -- which is, at best,
only partially correct. The subtle but crucial (I'd say
incontrovertible) evidence for their true relative chronology comes from
Kluge's Law, which is rarely if ever mentioned in handbooks. It's so
little known that even such an incisive debunker of linguistic myths as
Roger Lass forgot to mention it in his discussion of GL vs. VL
(_Historical Linguistics and Language Change_, section 5.4). It may seem
didactically convenient to present the "laws" as sequentially arranged
monolithic blocks (and, needless to say, exclude anything that might
complicate the picture). In most cases students can painlessly derive
the correct output, which eventually convinces them that they've been
taught the Truth. I used to follow the easy handbook path myself (we are
all sinners) but I no longer do so. The harder way is more instructive
and at the same time more satisfying -- also to the students, I hope.
Piotr