Concerning the plural problem in Serbian

Thu Apr 25 07:30:56 UTC 1996

> If this story is a workable account, the examples of plural collectives
> cited by Mr. Shipka should not have any grammatical status; that is,
> they should be rare and idiomatic.  I am unfamiliar with them though
> I'm not a native speaker.  Perhaps we could get some idea of what they
> represent from other SEELANGers.   My hypothesis of morphological
> categories does not allow them (unless they could be interpreted as
> nouns whose reference is Number-relevant but neither singular nor plural,
> the replacement of Dual as [-Plural, -Singular].

I am not the other SEELANGer :) but here are some new information.

To make our discussion more precise, let me add that I was
discussing the status of the -ad suppletive plural within
certain declension type ("II vrsta, pros"irenje na -et-).
Now, within this category, the -ad plural is dominant. There are
no reason whatsoever to consider them rare within this declension
type, but -et- stem declension itself is rather rare (which can be
seen from the number of examples extracted from the NT dictionary).

This is observed in a number of grammars (I quoted Stiber
last time), and can be easily confirmed with native speakers.

I extracted neuter -et- stem nouns (or historically -nt-/-e,t-)
from the NeuroTran (30 000 S-Cr entries) dictionary, and asked
three native speakers, if they can have -ad plural. This is
the list and the results. (+ = can have -ad plural)

bure,eta n; +
burence,eta n; +
}ebe,eta n; +
}ure,eta n; +
dete,eta n;
detence,eta n; +?
dijete,eta n;
djete{ce,eta n; +?
dojen~e,eta n; +
dugme,eta n; +
|ubre,eta n; +
gu{~e,eta n; +
jagnje,eta n; +
janje,eta n; +
jaje,eta n;
jezerce,eta n; +?
june,eta n; +
kljuse,eta n; +
kopile,eta n; +
ku~e,eta n; +
kube,eta n;
ma~e,eta n; +
mladun~e,eta n; +
nedono{~e,eta n; +
nevina{ce,eta +
pa~e,eta n; +
par~e,eta n; +
pile,eta n +
prase,eta n; +
puce,eta n; +
sandu~e,eta n; +?
seoce,eta n;
sir}e,eta n;
siro~e,eta n; +
{tene,eta n; +
tele,eta n; +
tuce,eta n; +?
unu~e,eta n; +
u`e,eta n; +
zamor~e,eta n; +
zrnce,eta n; +
zvonce,eta n; +?
`drebe,eta n; +
`drijebe,eta n; +

44 cases 5 does not have -ad plural, with 6 there were doubts =>
At least 75% of the cases have -ad plural

Moreover, in some cases, such as nedono{~e, this
is the only solution.

This list does not contain the words which are not in the
dictionary, such as ethnonyms: Cigan~e, Crn~e..., culture
bound words: kum~e, the words on the verge of being
dialectisms: pa{~e, and some other words, such as
parampar~e.

This last example shows that -ad form is also used in
phraseology: "razbiti se na parampar~ad", while
*"razbiti se na parampar~eta" is not grammatical.

To really prove the status of the suppletive plural in
this category, one should have:

1. more than 50 native speakers as her subjects,
2. questionnaire with different assignments (such as
   converting sentences to plural, choosing one from
   several forms, etc.)
3. SPSS or another package to do the statistics

This can also be checked by searching these -ad plural
nouns in a text corpus (prof. Moerk's corpus is big
enough for these purposes).

When you have to solve concrete problems, it does not make
any sense to consider this plural idiosyncratic one. For
example, in the rules NeuroTran uses to generate S-Cr
paradigms, called Minimal Information Grammar, the rule goes
(using the entries as they are given above but without + and
?):

SCR PARA *e,eta n =>        ! if you find lexical entry ending in e,eta n
 NOUN;NEUTER;O1=(1->','-1); ! take the part to the comma and
 SINGULAR;                  ! get rid of the last character (you get, for
  NOM=O1+e;                 ! example (mladun~e)-e = mladun~), and finally,
  GEN=O1+eta;               ! add the following endings
  DAT=O1+etu;
  ACC=O1+e;
  VOC=O1+e;
  INS=O1+etom;
  LOC=O1+etu;
 PLURAL;
  NOM=O1+ad;
  GEN=O1+adi;
  DAT=O1+adima;
  ACC=O1+ad;
  VOC=O1+adi;
  INS=O1+adima;
  LOC=O1+adima

The point is that least information is needed when you formalize
this declension type like default. All other cases are hadeled using
additional labels in the dictionary entries. This way you have to
give additional information to 25% rather than 75%, of course, if
your goal is to generate one grammatical plural form. More information
about MIG and NT at:

http://www.amu.edu.pl/~sipkadan/mig.htm
http://www.amu.edu.pl/~sipkadan/nt.htm

BTW, I can extract different word lists, like the nouns with longer
ov/ev plural, the verbs of the nu/ne type, words with suffix
-acija, words connected with computer science - all for S-Cr, and,
of course, for all other flection types, affixes, and usage fields.
If you are doing research in S-Cr, let me know at my private address.

Sincerely,

Danko Sipka
-------------------------------------------------------------------------
Dr. Danko Sipka
Slavic Department, Adam Mickiewicz University, Poznan
Institute of Philosophy and Sociology, Polish Academy of Sciences, Warsaw
Translation Experts Ltd, Polish Division, Poznan
-------------------------------------------------------------------------
e-mail:     sipkadan at hum.amu.edu.pl or sipkadan at plpuam11.amu.edu.pl
www:        http://www.amu.edu.pl/~sipkadan/ja.htm
phone/fax:  ++48-61-535-143
mail:       Strzelecka 50 m. 6, PL-61-846 Poznan, Poland
-------------------------------------------------------------------------