non-classifier languages without plural inflexion

[Alan R. King]

Edith Moravcsik says, in reply to Bingfu's query on languages with no
classifiers and no obligatory number marking on nouns:

>Hungarian is a language which has no numeral classifiers to speak of
>and, even though it does have a nominal plural affix, nouns
>do not ALWAYS receive this affix when they have a plural referent.
[my emphasis added]

There are numerous languages that would fill this description, but I would
have assumed that this is not quite the same thing as what Bingfu had asked
for (maybe Bingfu should say, though!).  The point I would centre on is
that in a language such as Hungarian, nouns do in general, obligatorily,
get marked grammatically for number; the cases in which this marking does
not occur comprise only a strictly limited subset of (syntactically,
lexically...) definable instances, e.g. when quantified by a numeral.  I
believe Hungarian contrasts significantly in this respect with Aymara
(which I gave earlier as an example): I think that in Aymara plural marking
is *generally* optional, whereas in Hungarian it is *generally* obligatory.

(I beg to be corrected if I'm wrong; neither of these languages is
particularly well known to me, and mistakes do happen. I also understand
that Aymara may be changing in this respect and that today, quite possibly
due to Spanish influence, nominal number marking is becoming more frequent
than it was originally.)

I can think of several languages that fit the same overall description as
Hungarian in this respect; two European ones that immediately spring to my
mind (and which I know far more intimately than Aymara or Hungarian) are
Basque and Welsh.  In both of these (which I should mention do not show any
generalized use of numeral classifiers), nouns or at least noun phrases are
in *unmarked* circumstances consistently and obligatorily marked for
number, i.e. are grammatically plural when their reference is plural.  In
the case of Welsh number marking is inflexional and attaches to the noun:

dyn 'man'
man

dyn-ion 'men'
man-PLU

In Basque things are rather different: it is the NP as a whole that
receives a plural marker, which is realized as a morphological element
fused with the determiner (the majority of NPs must contain one determiner):

gizon-a 'man'
man-DET

gizon-ak 'men'
man-DET.PLU

gizon hura 'that man'
man that

gizon haiek 'those men'
man those

Note:  Since the unmarked (mostly obligatory) determiner is singular -a,
plural -ak, Basque grammars often treat -ak simply as the plural morpheme.

In Welsh, the chief exception is that nouns quantified by a numeral retain
the singular form:

tri dyn 'three men'
three man

In Basque, some determiners are inherently (semantically) plural per se
without containing a number morpheme.  (Actually we're talking about
quantifiers here; it is convenient for the analysis of Basque grammar to
treat quantifiers as a subclass of determiners.)  In NPs containing such
determiners, the reference will be plural but there will be no visible
plural marker.  Inherently plural determiners are more or less the ones we
would expect: numerals above one, and quantifiers meaning 'much/many' and
so on (whose reference is plural when applied to count nouns, not to mass
nouns, of course):

hiru gizon 'three men'
three=DET man

gizon asko 'many men'
man much/many

ur asko 'much water'
water much/many

But since in the unmarked case an NP has the default determiner -a/-ak, and
since this (and some other determiners) are obligatorily number-marked, I
would maintain that, in general, in Basque noun(-phrase)s are
systematically marked for number.

As for Welsh, it is only nouns with a numeral that stay singular when their
reference is plural.  Unless, that is, we want to talk about a subset of
Welsh nouns whose plural is morphologically unmarked while the *singular*
is marked inflexionally, such as

madarch 'mushrooms'
mushroom

madarch-en 'mushroom'
mushroom-SIN

But this does not in the least affect the point that nouns are obligatorily
marked for number, it is only of morphological interest.

A further language that behaves something like this, perhaps tending
slightly closer to what Bingfu (I think) asks for without quite getting
there, is Hawaiian.  Nouns are not marked inflexionally for number (or
anything else), but as in Basque, the general rule is that NPs must contain
a determiner, and again there is a default determiner (which in both Basque
and Hawaiian grammars is typically designated as definite article, in my
opinion a not-entirely-satisfactory projection of a mainstream
west-European linguistic concept), and once again this default determiner
has a singular and a plural form (singular KA or KE, plural NA:):

'o ke keiki '(the) child'
PARTICLE DET child

'o na: keiki 'children'
PARTICLE DET.PLU child

(I've used a different noun, because _kanaka_ 'man, person' and a handful
of other nouns denoting animates are exceptions, and actually do vary
inflexionally for number!)

When any other determiner is present, plurality is marked in a different
way, by a kind of quantifier morpheme which immediately precedes the noun:

'o keia keiki 'this child'
PARTICLE this child

'o keia mau keiki 'these children'
PARTICLE this PLU child

On the whole, although number marking is not quite as systematic as in the
European languages seen, it is integrated into the grammar and regularly
reflects the singular or plural reference of the NP.  Regularly, yes, but
not inevitably; I quote Emily A. Hawkins:

"The indication of singular and plural is not as frequent in Hawaiian as in
English.  Whereas, while [sic] in English the plural must be indicated, in
Hawaiian the plurality is dictated by the context.  Pluralizers eliminate
the possible interpretation of singularity, but their lack does not
eliminate a plural reading." (p. 9)

If my understanding is correct, in this paragraph we could substitute
"Aymara" for "Hawaiian" and it would remain true.  That doesn't necessarily
mean that the extent of the use of pluralizers coincides in the two
languages, of course.  In the Hawaiian texts I've seen, it is my impression
that grammatically explicit pluralization still is the rule rather than the
exception, though not so much of a "rule" as in Basque, Welsh, and
presumably Hungarian.

Perhaps Juergen (hi, Juergen!) can say something about the partly but not
entirely similar case of Tongan, a language related to Hawaiian, in this
respect?

Summing up: there seems to be a continuum relating languages in which
nominal number is "grammatical" to different degrees, and which in
consequence may or may not meet Bingfu's criteria (assuming they lack
numeral classifiers) to different degrees.  Romance and Germanic languages
stand at one end of the continuum, with VERY "grammatical" number.  A small
distance away, but still I should think well over towards the
Romance/Germanic end of the gradient, would be languages like Hungarian,
Welsh and Basque.  Followed then by Hawaiian, and Aymara, and so on.  Even
in these last, I would say number remains PRETTY grammatical, all told.
This is still a long shot from languages WITHOUT systematic number marking
on NPs.  Fijian is one, I believe (it has it in the subject person/number
indices that must accompany the verb, though) - and the use of numeral
classifiers is not very prominent.  Try Fijian, Bingfu!  (Why didn't I
think of this before?)

Source:
Emily A. Hawkins, _Pedagogical grammar of Hawaiian: recurrent problems_,
University of Hawai'i, 1982.