numerals and number markers
Edith A Moravcsik
edith at CSD.UWM.EDU
Wed Feb 11 21:14:40 UTC 1998
Dear ALT Colleagues,
Recently, the four of us - Grev Corbett, Alan Dench, David Gil, and
Edith Moravcsik - became engaged in an e-mail discussion
about criteria that may serve to differentiate between
numerals and quantifiers, on the one hand (e.g. "two" or "many")
and grammatical number markers, on the other (i.e., dual, trial,
or plural markers). The discussion was triggered by Grev's paper
on grammatical number given at the ALT-II meeting in Eugene, Oregon,
this past September and started by David, who called attention
to some constructions from English and Hebrew which seemed to
be borderline cases between involving a numeral/quantifier
and a grammatical number marker. For example:
English: we two
we three
Hebrew: SloS-t-enu "three-CONSTR-1:PL" 'we three'
Further examples from other languages:
Hungarian: (mi-)kett-en "we-two-AFF" 'we two'
(mi-)ketto"-nk-et "we-two-lPL-ACC" 'us two'
Yinhawangka: ngunha-kutharra-ma "that-two-EMPH" 'those two'
Breton: hon-daou "us/our-two" 'we two'
ho-tri "us/our-three" 'we three'
(Note: re Hebrew: hyphen placement is a rough approximation
re Hungarian: the diacritic " should go on the preceding
vowel)
In each case, the question is the same: are these words number-marked
forms of pronouns or are they numerals showing person and number?
The conclusion that we tentatively settled on was the conventional
one: free versus bound status. That is:
a/ BOUNDNESS CRITERION
If the quantifying formative is a free form, it is a numeral or
quantifier; if the formative is an affix, it is a
grammatical number marker.
By this criterion, all the forms above are inflected numerals
rather than number-marked pronouns since the quantifying formative
they include is in each case a free form. Note that the
boundness criterion does not always provide for a clear cut: it
leaves open the analysis of forms like Tagalog _mga_ and that of
"plural words" (Dryer 1989).
In the course of the discussion, we tried other possible criteria
as well. These were the following:
b/ CARDINALITY
Does the quantifying formative express cardinality or not?
If it does, it is a numeral; if it does not, it is
a grammatical number marker.
This is clearly in conflict with the boundness criterion:
formatives expressing cardinality may be inflections
(such as dual or trial markers) and formatives without
cardinality may be free forms (such as English _few_ or
_many_).
c/ LOWER NUMBERS?
Assuming that cardinality (as in b/) is not the right criterion
so that both numerals/quantifiers and grammatical number
markers may or may not express cardinality, a partially
applicable criterion may be whether a quantifying formative
that does have cardinality is or is not restricted to lower
numbers. If it is, the formative is a grammatical number
marker; if it is unrestricted in terms of how high the number
can be, it is a numeral.
This criterion does not support the boundness criterion for the
English, Hebrew and Hungarian constructions cited above. These
constructions involve free-form numerals; yet, they must be
lower numerals. Also, there are languages where free-form
numerals never go above 5 or 10.
d/ HEAD OR DEPENDENT?
Is the quantifying formative a head or a dependent? If it is a
head, it is a numeral or quantifier; if it is a dependent, it
is a grammatical number marker.
This is again in conflict with the boundness criterion. In the
Hungarian examples above, the numeral is a free form and the
pronominal element is an affix and thus the boundness
criterion would say these words are inflected numerals, But,
interestingly, if the word serves as a controller of agreement
(with the verb agreeing with it as subject or as object, or the
reflexive pronoun agreeing with it as subject), the verb and
the reflexive show the person and number of the pronoun, rather
than being in the third person plural as one would expect if
the numeral were the controller and thus the head.
e/ STACKING
If the construction involves more than one quantifying
formative, is their interpretation flat or hierarchical? If it
is flat, such as in English _two eyes_, one of the quantifying
formatives is a grammatical number marker and the other, a
numeral or quantifier. If the interpetation is hierarchical
(i.e., the two multiply each other, such as in Breton _daou-
lagad-ou`_ "DUAL-eye-PLURAL" meaning 'pairs of eyes'; for
other similar cases, cf. Japanese), both are
numerals/quantifiers.
This criterion supports the boundness criterion in some
instences. For example, English _two eyes_ has two quantifying
elements (_two_ and _-s_) and a flat interpretation; and it is
true that one of the quantifying elements is free and the
other is bound. Thus _two_ classified as a numeral and _-s_
as a grammatical number marker is consistent with both
criteria.
In other instances, however, the criteria of stacking and
boundness are in conflict. For example, in the above-cited
expression of Breton (_daou-lagad-ou`_ "DUAL-eye-PLURAL" 'pairs
of eyes'), stacking would rule both the dual and the plural
element numerals since the interpretation is hierarchical; but
boundness would rule both as grammatical number markers since
both are affixes.
Another problematic case is Georgian. The following Georgian
phrase is ambiguous:
xut-i or-or-i vaSl-i
ten-ABS two-DISTR-ABS apple-ABS"
The above phrase may be interpreted either hierarchically or
flatly. The flat interpretation is 'ten apples in groups of
two' - a total of ten apples; the hierarchical (or stacked)
interpretation is 'ten sets of pairs of apples' - a total of at
least twenty apples. As far as the status of the two
quantifying elements _xut-i_ 'ten-ABS' and _or-or-i_ "two-
DISTR-ABS" is concerned, the stacking criterion makes different
predictions depending on the interpretation. Under the
hierarchical interpretation, both would have to be numerals and
this is so far in line with the boudness criterion since both
are free forms. Under the flat interpretation, however, the
stacking criterion would class either _or-or-i_ or _xut-i_
'ten' as a grammatical number marker even though they are both
words.
f/ ANIMACY HIERARCHY
Do the cooccurrence constraints between the quantifying
formative and the noun or pronoun it goes with follow
the version of the animacy hierarchy as proposed by T. Cedric
Smith-Stark (1974) for predicting the distribution of
grammatical plural markers? If it does, it is a grammatical
number marker; if it does not, it is a numeral or quantifier.
This criterion conflicts, at least in some cases, with the
boundness criterion. For example, the English, Hebrew and
Hungarian forms cited above include a free numeral but the
construction is nonetheless restricted to pronouns - a
constraint that the Smith-Stark hierarchy would predict to hold
for grammatical number markers rather than for numerals.
At this point, our discussion petered out. There are
nonetheless some further questions that we see as wide open:
A. Are the criteria listed above reasonable to begin with?
B. The criteria clearly do not always coincide. To exactly what
extent do they converge and diverge? Are there any
implicational relations that could be set up as holding
among them?
C. Are there any other criteria that one could rely on to
tell apart grammatical number markers and
numerals/quantifiers? (For example, linear order relative
to the noun or pronoun?)
D. Is it necessary to draw a distinction between quantifiers/
numerals and number markers to begin with?
We hope that you will want to offer some ideas on these issues
or ask us any questions about the above.
Grev Corbett (g.corbett at surrey.ac.uk)
Alan Dench (adench at cyllene.uwa.edu.au (Alan has
not had a chance to approve of this summary)
David Gil (dgil at strauss.udel.edu)
Edith Moravcsik (edith at uwm.edu) (in charge of
the write-up)
References:
Dryer, Matthew. l989. "Plural words." _Linguistics_ 27. 865-
895.
Smith-Stark, T. Cedric. l974. "The plurality split." Papers
from the Tenth Regional Meeting, Chicago Linguistic Society.
657-671. Chicago: Chicago Linguistic Society.
More information about the Lingtyp
mailing list