willemvermeer wrote:
> --- In cybalist@yahoogroups.com, "Grzegorz Jagodzinski"
> <grzegorj2000@...> wrote:
>
> [On the Progressive Palatalization]
>
>> But why this sound rule is so difficult? Because there are plenty of
>> proposed conditions (and counterexamples). We should answer the
>> following questions:
>
>
>> 1) how could have existed and functioned so multi-limited phonetic
> rule?
>
> In the traditional formulation it is a very ordinary phonetic rule,
> to wit a progressive palatalization of velars preceded by high front
> vowels unless followed by a consonant or a high rounded vowel.
I wish I had more time for this discussion. Anyway, there are phonetic rules
which are unexceptional or nearly unexceptional. The 3rd Slavic
palatalization, or the progressive palatalization, does NOT belong to this
group. All the rest is the question of interpretation and terminology.
Neogrammarians believe that all phonetic rules (called "laws" by them) are
unexceptional - but sometimes analogy can cause retreating the change which
once occured due to the rule. In some other instances they believe that
analogy can cause the sound change to occur even when there are no proper
conditions for the change. You have the right to think that all is right
here. But I have right to term white white and black black - and that's why
I term the 3rd palatalization phonetically irregular.
In my terminology, "regular" means "with no exceptions", and it is not
important whether these exceptions are caused by analogy or by something
else. And the concept of "regularity" is not "Boolean" for me: in other
words,
a change can be almost completely regular (it occured in almost all
situations when it was expected), it can be less regular, or it can occur
only in some instances without clear conditions. And I consider the
neogrammarian point of view worse for a number of reasons:
1) We have the full view of the change only in rare instances. So, we just
cannot say that, for instance, the prograssive palatalization in Slavic had
occured in all words with 100% regularity, and only next analogy blurred
that beautiful picture. And only in such instances we would have the right
saying that the change was regular but next analogy occur. Neogrammarians
must assume such imaginary scenario, I need not it for anything.
2) Of course, Neogrammarians can also say that analogy never allowed some
forms to exist - or that it stopped the rule to function in some instances.
They would say that their unexceptional rule SHOULD function but analogy
prevented it. My point allows to avoid such anthropomorphisms. A rule
"should" nothing. It just functions or it does not. And its limits can be
various: "natural" limitations (the ones given in the formulation of the
rule), analogy, frequency, even a blind chance sometimes.
3) I believe it is commonly known that limits of most language phenomena
need not form strict dialectal borders. According to Neogrammarians, one
village changed its dialect due to a rule they analyse, and the neighbour
village, say 5 km from the former, did not. And no intermediate states are
possible - but rules should be unexceptional and "Boolean" (true or false,
and tertium non datur). So, no intermediate states should be possible - but
they are! Take the second Germanic consonantal shift, or the Slavic
metathesis (practically in all instances in Polish, rarely instances in
Polabian, and 50:50 in Kashubian). This picture is in better accordance with
the model with various levels of regularity than with the model with only
fully regular changes.
4) Neogrammarians' "laws" are PHONETIC. It means that they should describe
sound changes, and their conditions must have phonetic nature as well. I do
not deny that there exist clearly phonetic rules. But I can also see
existence of sound changes which are conditioned by morphology. Such
morphologic conditions can also be less or more regular. For example the
final *-os yielded Slavic -U or -o, and it depended on its function (cf. -U
in nom.sg. masc. but -o in nom.acc.sg. neutr.) rather than the phonetic
environment. The explanation I follow is better than analogy: notice also
*-om > -U in acc. masc. and -o in acc. neutr - what is regular and what
analogous here? And to what is the analogy here? The modern Polish 1 pl.
verbal ending -my instead of old Polish -m (< -mU) is analogous to the
pronoun my 'we', but what about Ukrainian -mo? Analogy? Or maybe a regular
development of *-mos? If yes, -mU cannot be analogous (to what?). Instead, I
see no place for analogy here and I prefer to explain both endings as the
results of clearly morphologic, non-Neogrammarian, rules. Development of the
final *-os and *-om is similar to the 3rd palatalization: it is limited both
territorially and functionally.
5) Followers of neogrammarian ideas waste their time and invent more and
more complicated conditions for these of their unexceptional rules which
have exceptions. Each theory is good as long as it allows to foresee
something. Some artificial limitations of phonetic rules formulated by
Neogrammarians are just the art for the art, or making metaphysics, and they
lack any heuristic value. Let's take the Slavic progressive palatalization
once again, in the shape you term traditional (and which is unknown to
authors of many traditional academic books). Indeed, the formulation lets us
understand such forms as kUne,gynji, lIgUkU, gen.sg. otIca, the masculine
formant -ikU, fem. -ica etc., but it does not allow to understand why
nom.sg. otIcI, gen. sg. masc. -ika, etc. Of course, one can always say that
these "irregularities" are caused by analogy - but no one can say why in
some instances the analogy caused changing nominative after genitive, and in
other instances - genitive after nominative. So, if blind and unpredictable
analogy is called every time when facts contradict the unexceptional rule,
we must assert that the rule is of no value. In other words: if facts are
different than foretold, we have an ace hidden in our sleeve, and we call
the ace analogy. Maybe even the more exception the better for the rule? But
some people say so... I do not want to deal with aces in sleeve, so I do not
like unexceptional rules and I do not like wasting time to formulate them.
Instead, I prefer the simplest possible rule of the third palatalization:
*k, *g, *x > c', 3', s' when after I, i, eN (and possibly R'), without no
further conditions but with the notice that it was not an unexceptional
rule.
The same formulation is applied in many academic books, including OCS
student's books and other materials for historical grammars used in Polish
universities. Other people may like such an approach or not, they may even
hate it and present their personal opinions instead - perhaps because thay
have no other ideas for topics of their scientific investigation.
Please, do not accuse me of such an opinion, I only support views of some
other people, keeping a level head, like professor Man'czak, a famous Polish
Romancist. Instead of concocting newer and newer explanations to save
unexceptional character of the rule, I have just found as follows:
1) sometimes it occurs, and sometimes it does not, independently of what
consonant follows the velar (for example, IgU in lIgUkU but IcI < IkU in
otIcI, -ika ~ -ica etc.),
2) but when it have already occured, it functions in all forms of the lexeme
(contrary to regular 1st and 2nd palatalization, it does not cause consonant
alternations in inflexion),
3) it does not occur when it would lead to a new morpheme in a limited
number of words (but it can occur when the morpheme would occur in many
words, cf. -IkU ~ -IcI, -ika ~ -ica etc.).
And I cannot find even one instance in which the Neogrammarian unexceptional
rule (with the limitations) would be able to predict a form while my
soberminded attempt would not. So I see no reason to think the Neogrammarian
rule is better.
>> In other words: self-protection with analogy is only a desperate
>> trial to
>> save neogrammarian ideas, nothing more. In order to admit that the
>> 3rd
>> palatalization was a regular process we would have to know that it
>> occured in all instances where it should occur, and only next some
>> of those instances were retracted because of analogy. But we have
>> not such knowledge.
> It is normal methodology to regard cases that can easily be the
> consequence of analogical levelling as uninformative.
But when _most_ instances can be the consequence of analogical levelling and
when this levelling can be both restricting (no change when expected) and
extending (the change occured even if not expected), the rule itself is of
really no value. Any theory is useful only when it can be applied to _most_
instances. If most examples do not supply the theory without clear rules,
the theory should be rejected. And analogy functions without clear rules. I
repeat: it can sometimes restrict and sometimes extend the rule.
>> Which is more, according to formulated rules, the instances like
> otIcI <
>> *otIkU should not show the 3rd palatalization (before -U). Why did
> it happen
>> in otIcI but not in numerous formations with -ikU?
> Because the *i in -ikU reflects earlier *ei, which does not trigger
> palatalization. This is standard knowledge. The examples are in the
> second edition of Meillet's "Le slave commun" (1934).
Once again, I do need the literature written a century ago to state that the
feminine counterpart of -ikU is -ica not **-ika. So, even if -ikU < *-eikos
then -ica < *-eika: Which one is irregular and why?
>> According to your
>> argumentation, analogy (as if) caused the 3rd palatalization to
> occur
>> despite of the presence of -U.
> No, first it is not "my" argumentation, but mainstream argumentation
> since 1910 or earlier.
... and rejected now, except authors who still believe in Neogrammarian
ideas, like really unexceptional phonetic rules.
> But what is much more important, the formulation distorts the
> traditional view of what constitutes morphological analogy. It is not
> a matter of the Progressive Palatalization "occurring" in examples
> like *otIkU, but of the stem form *otIc-, which was found in such
> forms as Gsg *otIca, being subsequently (quite possibly several
> generations after the Progressive Palatalization had taken place)
> generalized to those cases in which the stem form *otIk- was
> phonologically regular, it is a matter of replacement resulting in
> simplification of the grammar.
A clear invention, not based on facts. And the fact is that regressive
palatalizations caused stem alternations. The explanation that the
progressive palatalization underwent analogy in order not to cause stem
alternations is not convincible then.
>> A tiny number of forms means a tiny number of pieces of evidence.
> Thus, we
>> have no base to assert that the 3rd palatalization was in keeping
> with the
>> thesis of regular character of phonetic processes.
> I'm inclined to agree. If somebody would stand up and launch
> Neogrammarian thinking on the basis solely of the Progressive
> Palatalization of Slavic, he would not have a particularly convincing
> case, but the discussion was about a different matter: is the
> Progressive Palatalization of Slavic an obvious example of a set of
> data Neogrammarian method is unable to handle?
And the answer is that it is such an example. Perhaps Neogrammarians think
that their methods can handle the example but it is not true like I showed
above. Neogrammarian methods are not acceptable: "we rejoice when facts are
in accordance with our rules - but when they are not, we have an ace called
analogy, so we need not be afraid of anything". If somebody accept such a
science, please do, most certainly. But I will not follow such ideas.
>>> The assumption that *y blocked the Progressive Palatalization is
>>> based on the retained velar in *kUneNgyn/i 'princess, queen' (not
> in
>>> OCS, but early attestations in Old Russian and elsewhere), a
>>> derivation from _kUneNdzI_ 'prince, king'. It is the only example
> of
>>> a retained velar in this stem and cannot conceivably be
> analogical.
>>> The first to see this was Josef Zubaty/ (Sborni/k filologicky/ 1,
>>> 1910, 150-153). To the best of my knowledge it has never been
>>> questioned since.
> Then Grz wrote:
>> A reliable rule (or a restriction of a rule) based on _one_
> example? It is
>> really funny, just like you have stated above.
> Yet it is ordinary Neogrammarian method.
And this is the reason for which I call the 3rd palatalization a typical
example of irregular change. I just do not accept such methods.
> For some reason outsiders
> and beginners are inclined to assume that Neogrammarian thinking
> consists exclusively of the concept of sound law plus a number of
> fudge factors. One doesn't have to read a lot of Meillet and Pedersen
> to see that that is a caricature of what really goes on.
> Numbers just aren't important here. What is important is the way the
> relevant evidence is accounted for. In the case at hand, we have a
> number of cases of modified velars followed by the reflex of *y plus
> a single case of a retained velar in the same position. The modified
> velars can all be easily explained as a trivial consequence of
> analogical levelling whereas the single retained velar, which takes
> part in an alternation hence _has_ to be explained, cannot be so
> explained.
And you are simply wrong in this point, as I have already stated. The only
example of kUne,gynji can be easily explained as a result of levelling - we
do not know any **-inji formations in Slavic. You seem to reject, for
unknown reason, the word-formation levelling, and you seem to permit only
flexional levelling. Facts contradict such an attempt: in regressive
palatalization analogy did not cause any flexional levelling. So, the belief
that unexpected modification (or: no expected modification) being a
conseqence of analogical levelling in inflexion is trivial - is motivated
with nothing.
> So if you assume the retained velar was retained because
> of the following *y, you account for all of the evidence. If on the
> other hand you assume that velars followed by *y were palatalized,
> you leave part of the evidence unaccounted for.
> This isn't all, of course. For one thing, [u:] (the earlier stage of
> *y) is just the kind of vowel to block a palatalization, so the
> restriction is phonetically plausible. For another, the assumption
> that the short counterpart of [u:] ([u], i.e. attested U] also
> blocked palatalization accounts for some evidence too, so the
> formulation can be generalized to the statement that high rounded
> vowels blocked palatalization.
Perhaps, but do not forget that the 3rd palatalization stays irregular
indepentently on whether you accept the condition of u-blocking or not. And
which is more, if you assume the blocking, you cannot explain such instances
like otIcI < *otIkU. The explanation that other casual forms caused
palatalization is implausible because such a process is not known for other
phonetic changes in Proto-Slavic. All known analogical levellings in
inflexion are relatively very young and seem to occur especially in the
vicinity of dialects where the 2nd palatalization did not occur (Russian
levellings and no palatalization in Novgorod).
> But even in the absence of contributing factors the retained velar of
> *kUneNgyn/i would be taken extremely seriously. To the best of my
> knowledge it has not been questioned since it was first brought into
> the discussion by Zubatý, despite the sometimes acrimonious character
> of the debate.
I have already written what authors of many books think of the 3rd
palatalization. They just ignore both the kUne,gynji example and the
u-blocking. And they are right... as I explained above.
> Then Grz wrote:
>
>
>> But it is methodologically
>> incorrect, and those who believe in such a rule are also believers
> in
>> neogrammarian-like exceptionless processes. It is hard to discuss
> with
>> beliefs, especially in the instance when rules based on one example
> are
>> taken into consideration. I cannot imagine that other branches of
>> knowledge would apply such one-example rules (and term them "laws" in
> addition).
> The word "law" is an innocent piece of inherited terminology. Nothing
> would change if a different word would have been chosen (which, in
> hindsight, would have avoided a lot of misunderstandings).
Of course, but why not to use the right terminology "rule" instead of "law"?
Rules can be limited in space nad time - laws shouldn't be. And I am under
impression that many modern authors write such terms as "Grimm's law",
"Verner's law" etc. in quotation marks, or term them just sound changes or
phonetic rules.
> As for the importance of numbers, linguists happen to have unhealthy
> traditions.
This is it! And that's why I do not use the term "phonetic law" but
"phonetic rule", and I accept some of them with reservation. If something is
unhealthy, it should be cured, shouldn't it?
>> We should accept another explanation here. There were plenty of
> forms with
>> the suffix *-ynji in Proto-Slavic but no forms with *-inji. If
>> *kUne,gynji had undergone the 3rd palatalization, it would have
>> yielded **kUne,3inji and it would have been the only word with
>> -inji. And this was the
> reason why the
>> change had not occured, not the presence of -y-.
> This is entirely speculative.
No way.
> There is nothing wrong with
> **kUneNdzinji, which would have lived on and given rise to Russian
> **knjazínja and Serbian **knèzinja etcetera.
It would have lived but it does not live... _This_ is speculation, not my
view.
> At the relevant stage of
> the development of Slavic just about all suffixes and endings
> occurred in variants differentiated by whether or not the preceding
> consonant was palatal, for instance the o-stem instrumental plural
> ending -y/-i or the suffix -yka/-ika found in vladyka vs. bliz^ika.
The former < *-u:ka: and the former < *-i:ka: They are two different
suffixes, perhaps partially mixed (masc. bliz^ika < ? blizj-yka, not
connected to feminine -ika, cf. also Pol. bliz'ni 'fellow creature' <
*blizI-njI). The problem is that *-inji is not known from Slavic (except of
new secondary formations like in Polish, due to the regular phonetic process
y > i in some environment).
>> And again, the reason why the 3rd palatalization did not occured
> here was
>> morphology (presence of -UkU and absence of -IcI among adjectives),
> not
>> phonetic environment. Let's imagine two following possibilities:
>> 1) there is a rule which functions in 2 instances and is stopped
>> because of analogy in 100 instances,
>> 2) there is a rule which functions in 100 instances and is stopped
>> because of analogy in 2 instances.
> Basically what you call morphological is exactly the same as what is
> called analogy in traditional neogrammarian methodology. It is a
> terminological difference of the same kind as "contrastive"
> vs. "distinctive".
The difference is that analogy is irregular by its nature. I search for
regularity in morphologic processes. For example, -IcI, -ikU, -ica are all
mophologic formations rather than phonetic (they are not results of regular
phonetic processes). The suffix -ika is also possible together with -ica but
(the former) without diminutive meaning (and it forms attributive masculina,
e.g. OCS bliz^ika - if not from *blizjyka -, or feminine plant names, ex.
osika 'asp(en)'). So, the meaning caused or did not cause the
palatalization, not phonetic conditions.
The same process - a regular morphologic rule - can be observed in many
other instances, perhaps in all languages. I'm going to recall only two of
them:
1) the difference between continuants of PIE nom.sg. *-o:(ns) in masculine
and feminine *-on- substantives (all phonetic explanations are highly
speculative),
2) the development of PIE *-os and *-om in Slavic (> -U in masc., > -o in
neutr. independently of whether it was *-os or *-om in PIE).
>> [...] and thus your explanation is not very convincing.
> I wish you wouldn't attribute mainstream thinking to me all the time.
I use "your" in the sense "the view you try to defend". And the same, I
write "my" thinking of what I defend. It does not mean that it is only mine.
I know authors who think similarly.
Grzegorz J.
___________________________________________________________
Yahoo! Messenger - NEW crystal clear PC to PC calling worldwide with voicemail
http://uk.messenger.yahoo.com