Distributive Reduplication in Telugu*
Rahul Balusu
New York University
1. Introduction
In Telugu, as in some of the other Dravidian languages like Kannada, Tamil and Malayalam,
reduplication of numerals gives rise to distributive readings. This is shown in (1) with the
three kinds of distributive readings that are possible for this sentence given in (2).
(1) ii pilla-lu renDu renDu kootu-lu-ni cuus-ee-ru
these kid-Pl 2 2 monkey-Pl-Acc see-Past-3PPl
lit. ‘these kids saw 2 2 monkeys’
(2) a. These kids each saw 2 monkeys. Participant key reading
b. These kids saw 2 monkeys in each time interval. Temporal key reading
c. These kids saw 2 monkeys in each location. Spatial key reading
The sentence in (1) contrasts with a similar sentence without the reduplicated numeral
as shown in (3) with the possible interpretations in (4).
(3) ii pilla-lu renDu kootu-lu-ni cuus-ee-ru
these kid-Pl 2 monkey-Pl-Acc see-Past-3PPl
These kids saw 2 monkeys.
(4) a. These kids saw 2 monkeys. (collective)
b. These kids saw 2 monkeys each. (distributive)
The above examples show that the reduplicated numeral construction always gives
rise to distributive readings, there are no collective readings of a reduplicated numeral
*
I would like to thank Anna Szabolcsi, Marcel den Dikken, Richard Kayne, David Beaver, Paul Elbourne,
Eytan Zweig, Tom Leu, Lisa Levinson, Oana Ciucivara, Laura Rimell, Liina Pylkkänen, Jason Shaw and the
audience at NELS 36 for all their very helpful comments and suggestions.
Balusu – to appear in NELS 36 Proceedings 2
construction like in (4a) and that it gives rise to two additional distributive interpretations that
are not present in the non reduplicated construction. The focus of this paper will be to explain
the obligatory distributivity that is associated with the reduplicated numeral ((henceforth
RedNum) construction and to account for the various additional distributive readings that are
possible in such constructions.
This paper shows that the reduplicated numeral is associated with a distributive
operator whose sorting key (Choe 1987) is always an event or an event aspect, i.e. that not
only the temporal and spatial but the participant key readings are also event key readings.
Moreover I propose that the construction is associated with a plurality requirement. This
links the proposal to the semantics of plurality as presented in Zweig (this volume). This is a
natural connection for two reasons. One is that reduplication as is known in the literature has
the semantic effects of pluralization and distributivity. The other reason is that the formal
semantics literature often describes the distributive readings of a predicate as the
pluralization of the predicate (Landman 1989, 2000). I propose that numeral reduplication
pluralizes the numeral phrase (5). The singular child in Malay and the numeral phrase 2
monkeys in Telugu undergo reduplication to form their respective plurals.
(5) Malay Telugu
anak ‘child renDu koolutu ‘2 monkeys’
anak anak plural of ‘child’ renDu renDu kootulu plural of ‘2 monkeys’
2. Terminology
Choe (1987) basically assumes that a distributive operator is a universal quantifier and has a
sorting key, i.e. the quantifier’s restriction, over which the distribution takes place and a
distributive share, i.e. the quantifier’s scope, that which is distributed. This is the definition
of the D-operator that is followed in this paper (6).
(6) D-operator sorting key distributive share
∀ set in restriction entities in scope
The members of the distributive share need not be exhaustively used up when being
distributed whereas the members of the sorting key need to exhaustively used up in being
distributed over. This is a key difference between the distributive share and the sorting key
that will be used in this paper to diagnose whether a particular DP contributes a sorting key
or is rather a distributive share that is distributed over some covert key.
As regards sorting keys, since events are always located in time and space, it is not
surprising that these two aspects or dimensions of the event are encoded in the semantics of
the event variable. As Link (1998) puts it “time stretches are assigned to events, and also
certain regions in space”. Link (1998) makes use of the temporal and spatial traces of events.
The temporal trace is the time period in which the event happens and the spatial trace is the
spatial region where the event happens.
Distributive Reduplication in Telugu 3
3. The Data and the Properties of RedNum
3.1 Intransitive constructions
The DP containing RedNum can be the only DP in the sentence (7).
(7) renDu renDu kootu-lu egir-i-niyyi
2 2 monkey-Pl jump-Past-3PPl
lit. ‘2 2 monkeys jumped’
This sentence can have one of the two interpretations (8).
(8) a. 2 monkeys jumped in each time interval. Temporal key reading
b. 2 monkeys jumped in each location. Spatial key reading
In (8a) there are 2 monkeys jumping each time, i.e. the monkeys are distributed in 2’s
over the temporal aspect of the event. Using the terminology of Choe (1987), the DP ‘2 2
monkeys’ is the distributive share. The temporal aspect of the event is the sorting key. I’ll
name this the ‘temporal key reading’. In (8b) there are 2 monkeys in each location, i.e. the
monkeys are distributed in 2’s over the spatial aspect of the event. Here again, the DP ‘2 2
monkeys’ is the distributive share. The sorting key in this reading is the spatial aspect of the
event; I’ll name it the ‘spatial key reading’.
3.2 Transitive construction with singular subjects
The DP containing RedNum can occur in a transitive construction with a singular DP as the
other argument (9).
(9) Raamu renDu renDu kootu-lu-ni cuus-ee-Du
Ram 2 2 monkey-Pl-Acc see-Past-3PSg
lit. ‘Ram saw 2 2 monkeys’
This sentence has two interpretations (10).
(10) a. Ram saw 2 monkeys in each time interval. Temporal key reading
b. Ram saw 2 monkeys in each location. Spatial key reading
Like the intransitive construction, the transitive construction with a singular DP has
two possible readings, the temporal key reading (10a), and the spatial key reading (10b).
3.3 Transitive construction with plural subjects
The reduplicated numeral can occur in a transitive construction with a plural DP as the other
argument (11).
Balusu – to appear in NELS 36 Proceedings 4
(11) pilla-lu1
renDu renDu kootu-lu-ni cuus-ee-ru
kid-Pl 2 2 monkey-Pl-Acc see-Past-3PPl
lit. ‘[the] kids saw 2 2 monkeys’
This sentence has three possible interpretations (12).
(12) a. The kids each saw 2 monkeys. Participant key reading
b. The kids saw 2 monkeys in each time interval. Temporal key reading
c. The kids saw 2 monkeys in each location. Spatial key reading
Along with the temporal key reading (12b) and the spatial key reading (12c) that exist
for the intransitive construction and the transitive constructions with a singular DP, the
transitive construction with a plural DP has an additional reading (12a) not seen in the earlier
two constructions, which I’ll name the ‘participant key reading’. In this reading, the DP
containing the plural acts as the sorting key, over which the RedNum phrase ‘2 2 monkeys’ is
distributed. Here a ‘participant’ in the event is acting as the sorting key, hence the term
‘participant key reading’.
3.4 Transitive construction with universal QP subjects
The RedNum DP can also occur in a construction with a universal quantifier (13).
(13) Prati pillavaaDu renDu renDu kootu-lu-ni cuus-ee-Du
Every kid 2 2 monkey-Pl-Acc see-Past-3PSg
lit. ‘Every kid saw 2 2 monkeys’
The sentence in (13) has three possible interpretations (14).
(14) a. Every kid saw 2 monkeys. Participant key reading
b. Every kid saw 2 monkeys in each time interval. Temporal key reading
c. Every kid saw 2 monkeys in each location. Spatial key reading
The readings in (14) are similar to the readings that exist for the construction with
plural DPs in (12) in that they have the same types of sorting keys.
3.5 Transitive construction with two RedNum phrases as subject and object
The two arguments in a transitive construction can both be RedNum phrases (15).
(15) iddaru iddaru pilla-lu naalugu naalugu kootu-lu-ni cuus-ee-ru
2 2 kid-Pl 4 4 monkey-Pl-Acc see-Past-3PPl
lit. ‘2 2 kids saw 4 4 monkeys’
1
Definite plurals in Telugu are not morphologically definite, i.e. there is nothing corresponding to a definite
article in the morphosyntax of the noun phrase. The indefinite and definite interpretations of the noun phrase
depend on the context. In a sentence like this the noun phrase is always interpreted as definite.
Distributive Reduplication in Telugu 5
The interpretations that are possible for such a construction are given in (16).
(16) a. 2 kids in each time interval saw 4 monkeys in each location.
Temporal & Spatial keys
b. 2 kids in each time interval saw 4 monkeys in each time interval.
Temporal & Temporal keys
c. 2 kids in each location saw 4 monkeys in each time interval.
Spatial & Temporal keys
d. 2 kids in each location saw 4 monkeys in each location.
Spatial & Spatial keys
Here there is no ‘participant key’ reading. This can be ascertained by checking
whether the members of either DP need to be exhaustively used up in the distributive
relation. Native speakers report that in none of the readings in (16) do either of the DPs have
to be exhaustively used up in the distributive relation. Therefore it can be concluded that
neither of them is the sorting key in any of the readings, and that the sentence does not have a
participant key reading. As there are two RedNum phrases, 2 distributions are possible. The
sorting keys are always either the temporal aspects of the event or the spatial aspects of the
event.
3.6 Constructions with RedNum in the subject position
In (17) is a sentence with the RedNum DP in subject position.
(17) iddaru iddaru pilla-lu kootu-lu-ni cuus-ee-ru
2 2 kid-Pl monkey-Pl-Acc see-Past-3PPl
lit. ‘2 2 kids saw [the] monkeys’
It can have three possible interpretations (18).
(18) a. Each of the monkeys was seen by 2 kids. Participant key reading
b. The monkeys were seen by 2 kids in each time interval. Temporal key reading
c. The monkeys were seen by 2 kids in each location. Spatial key reading
Like in (11), where RedNum was in the object position, there exist three readings, the
participant key reading (18a), the temporal key reading (18b), and the spatial key reading
(18c).
3.7 Descriptive generalizations
The descriptive generalizations that can be made from the above examples about the
behavior of RedNum are the following (19):
(19) a. In a transitive construction with a singular subject or when both the DPs in a
transitive construction have RedNum, and in the intransitive RedNum
Balusu – to appear in NELS 36 Proceedings 6
construction, only the spatial and temporal key readings are possible.
b. In a transitive construction with a plural or universal, along with the spatial and
temporal key readings, the participant key reading is available.
c. The RedNum DP can never be the sorting key.
d. RedNum can occur in both subject and object positions, and there is no obvious
subject/object asymmetry with respect to the possible range of readings.
The task of the analysis in this paper is to produce a unified account of this diverse
group of constructions involving RedNum, and to account for the readings that are possible
with each kind of construction.
3.8 RedNum vs. Binominal Each
As RedNum always forces a distributive interpretation, it is reminiscent of binominal each in
English. But there are major differences between them.
RedNum can occur with a singular DP, whereas binominal each cannot (20) (From
here on I will drop the actual Telugu sentences and just give the literal English glosses).
(20) a. ‘RAM SAW 2 2 MONKEYS’ = (9)
b.*John saw 2 monkeys each
RedNum can occur in a construction with a distributive universal but binominal each
cannot (21).
(21) a. ‘EVERY KID SAW 2 2 MONKEYS’ = (13)
b.*Every kid saw 2 monkeys each
The sorting key can be covert as in the temporal and spatial key readings or overt as
in the participant key reading in a RedNum construction with plurals and universals, as
shown earlier, whereas a binominal each construction only allows the participant key reading
(22).
(22) a. ‘[THE] KIDS SAW 2 2 MONKEYS’ = (11)
The kids each saw 2 monkeys. Participant key reading
The kids saw 2 monkeys in each time interval. Temporal key reading
The kids saw 2 monkeys in each location. Spatial key reading
b.The kids saw 2 monkeys each.
The kids each saw 2 monkeys. Participant key reading
*The kids saw 2 monkeys in each time interval. Temporal key reading
*The kids saw 2 monkeys in each location. Spatial key reading
As a result of these differences the analyses for binominal each won’t carry over to
RedNum. The analysis I propose for RedNum might be restricted to describe binominal each.
Distributive Reduplication in Telugu 7
4. Analysis
4.1 The analysis of Spatial and Temporal key readings
I’ll begin the analysis with the RedNum constructions that only have the temporal and spatial
key readings, the intransitive construction and the singular subject construction, repeated
here as (23) and (25).
(23) ‘2 2 MONKEYS JUMPED’ = (7)
(24) a. 2 monkeys jumped in each time interval. Temporal key reading
b. 2 monkeys jumped in each location. Spatial key reading
(25) ‘RAM SAW 2 2 MONKEYS’ = (9)
(26) a. Ram saw 2 monkeys in each time interval. Temporal key reading
b. Ram saw 2 monkeys in each location. Spatial key reading
I propose that RedNum has a D(istributivity)-operator associated with it and that takes
an event, or event-aspect as its argument. The temporal and the spatial key readings are
special cases of the event key reading.
But what do ‘each time interval’ and ‘each location’ refer to? After all time and space
are not atomic, they are not inherently chunked out into minimal units. They are mass-like.
The division of the spatial and temporal regions into units happens according to the context.
The units need not be of equal duration in the case of temporal regions or of equal
dimensions in the case of spatial regions (27).
(27) a. Temporal Division I
b. Temporal Division II
5 10 15 20 25 30
30252015105
Balusu – to appear in NELS 36 Proceedings 8
Imagine a situation where there are a number of monkeys in the monkey pavilion at a
zoo. Suppose every 5 minutes, there are 2 monkeys that jump (27a); and they can be next to
one another or very far from one another. Then a kid observing these monkeys can say ‘2 2
monkeys jumped’. He is chunking the temporal continuum into 5 minute intervals and it is
true that for each of these intervals there were 2 monkeys in it that jumped. Therefore the
sentence is felicitous. Now suppose that 2 monkeys jump up not at regular intervals but every
time that a bell is rung and the bell is rung irregularly (27b). In this situation too a kid
watching the monkeys could say ‘2 2 monkeys jumped’, this time chunking the temporal
region into intervals delineated by the ringing of the bell. So the measure using which the
temporal region is divided is based on some contextually salient parameter, and as long as
there are 2 monkeys that jumped in each interval of the time region divided up in this
fashion, the sentence is felicitous. The situation is similar for the spatial division as well.
Now suppose that all the monkeys’ foreheads were colored, such that for each color
there were 2 monkey of that color that jumped. A kid observing these monkeys could not
felicitously say ‘2 2 monkeys jumped’ by which he meant that 2 monkeys of each color
jumped. So a division of the jumping event based on kind or type is not possible. It is only
along the spatial or temporal dimensions that the event can be divided into intervals over
which the ‘monkey jumpings’ can be distributed.
According to the analysis so far the sentence in (23) ‘2 2 monkeys jumped’ will have
the interpretation of the form: There was an event e such that for every relevant part e’ of e,
two monkeys jumped in e’. Now what is every relevant part? I’ll be using the notion of a
partition (a set of non-overlapping subsets or parts) to model the relevant parts of e. The
figure in (28) is an example of a partition. The circle is the event e that is partitioned along
the spatial or the temporal dimension into a set of non-overlapping parts by a contextually
salient method of division as we just saw. Here a, b, c, d, f, g and h are the non-overlapping
parts or cells of the partition.
(28) π(e) = {a,b,c,d,f,g,h}
f igu re
The sentence in (23) is then given the (preliminary) form (29):
(29) ∃e ∃π(e) [ ∀e’∈π(e) ∃X[two_monkeys(X) ∧ jumped(X, e’) ]]
c
a
b
d
f g
h
Distributive Reduplication in Telugu 9
There exists an e such that there is a partition π of e such that 2 monkeys jumped in
each cell of the partition. Here e is a particular set of events. It is the parthood of the events
that I’m concerned with but I’ll be modeling it with a set. π is a contextually salient partition
of e. For the purposes of this paper I will not explicitly distinguish between the spatial and
temporal aspects.
4.2 Plurality requirement
The question arises whether the 2 monkeys that jump up per cell have to be different or not,
i.e. could the same 2 monkeys be doing the jumping every time.
Suppose there are 3 monkeys in the pavilion, but it is always the same 2 monkeys
which jump up when the bell rings. A kid observing the monkeys cannot felicitously say ‘2 2
monkeys jumped’ with the temporal key reading in mind, because it is the same 2 monkeys
that jumped up every time. But suppose there was a rule that if every time the bell rang, 2
monkeys jump up, then the kid can feed the monkeys. In such a case the kid can say to the
zoo keeper ‘2 2 monkeys jumped’ intending the temporal key reading, even in the above
situation when it was the same 2 monkeys that jumped each time that the bell rang. In this
context the identity of the monkeys is not relevant. This is similar to the felicity conditions
of the Event-related Reading (ER) of sentences like ‘Last year, 4000 ships passed through the
lock’ that Doetjes and Honcoop (1997) discuss. They suggest that a necessary condition for
the felicity of ER is that the identity of the ships be both irrelevant and easy to ignore.
Thus, I assume that the pairs of monkeys must be distinct or if not, their identity must
be irrelevant and easy to ignore. I will call this the plurality requirement: there needs to be
more than one pair of monkeys jumping in e. But the formula in (29) does not capture this.
The plurality requirement that we want to add should not be a conjunct because if the
same two monkeys jumped in every cell of the partition, (23) is not false, but is rather
infelicitous in some way. The plurality requirement is either a presupposition or an
implicature, it is clearly not a conjunct in the assertion. Zweig (this volume) argues that the
plurality requirement in English plurals is a conversational implicature. In this paper I will
not take a stand. I simply formulate the plurality requirement as an added condition. The
logical form of the sentence in (23) ‘2 2 monkeys jumped’ will now be (30):
(30) a. ∃e ∃π(e) [ ∀e’∈π(e) ∃X[two_monkeys(X) ∧ jumped(X, e’)]]
b. |{X: two_monkeys(X) ∧ jumped(X, e)}|>1
The content of (30b) is reminiscent of the implicature in Zweig’s proposal that uses
E-type anaphora to events: the cardinality of monkey pairs jumping in e is greater than one.
As will be shown the plurality requirement will also block the participant key reading
in a RedNum construction with a singular subject.
Balusu – to appear in NELS 36 Proceedings 10
4.3 The readings with Universals
4.3.1 Temporal and Spatial key readings - Easy
The RedNum construction with a universal quantifier is repeated in (31) with the possible
readings given in (32):
(31) ‘EVERY KID SAW 2 2 MONKEYS’ = (13)
(32) a. Every kid saw 2 monkeys. aa Participant key reading
b. Every kid saw 2 monkeys in each time interval. Temporal key reading
c. Every kid saw 2 monkeys in each location. Spatial key reading
In (32b) and in (32c), the temporal and the spatial key readings, there are two
distributions going on, one associated with the distributive universal quantifier ‘every kid’
and the other with the RedNum DP, with the temporal and spatial aspects of the event as the
sorting keys for the D-operator associated with RedNum in (32b) and (32c) respectively.
These readings can be analyzed like the earlier temporal and spatial key readings, by
using a partition of some event e for every kid. This is shown in (33):
(33) a. ∃E[∀y[kid(y) → ∃e∈E ∃π(e) [∀e’∈π(e) [∃X[two_monkeys(X) ∧
saw(y, X, e’) ]]]]]
b. |{X: two_monkeys(X) ∧ ∃y[kid(y)[saw(y, X, E)}|>1
There is an event E such that for each of the kids there is an event e which is a part of
E such that there is a partition of e such that the kid saw 2 monkeys in each cell of the
partition associated with him/her. When the partition is of the spatial domain, the spatial key
reading arises and when the partition is of the temporal domain the temporal key reading
arises. This analysis is able to account for the double distribution that we find in the temporal
key reading and the spatial key reading.
4.3.2 Participant key reading - A Problem: Redundancy of distribution
In (32a), the participant key reading, the universal QP, ‘every kid’ is apparently acting as the
sorting key. It was shown for the temporal and spatial key readings with a universal that an
additional distributive mechanism is contributed by the universal distributive quantifier
‘every’ besides the distribution associated with RedNum. But in (32a), there is only one
distribution going on. So for the participant key reading with universals there is a certain
redundancy of distribution.
What is puzzling is how the universal quantifier can act as the sorting key. The
universal quantifier already associates with its own distributive operator. This is a problem if
we take the view that the quantifier phrase ‘every kid’ is contributing the sorting key for the
D-operator associated with RedNum.
Distributive Reduplication in Telugu 11
4.3.3 Participant key reading - Changing the perspective on the sorting key
I propose is that the participant key readings are also event key readings. In these readings
also there is a partition, but the partition is a trivial partition where the whole event is the
single cell in the partition.
(34) The trivial partition: π(e) = {e}
Notice that a partition of a set A is a set B that contains non-overlapping subsets of A
and exhausts A. A is a subset of itself. Thus the trivial partition, with the whole event as the
single cell is as good as a partition which is non-trivial.
To analyze the participant key readings I’ll start with the construction with the
universal ‘every kid saw 2 2 monkeys’. This now gets the same interpretation as the temporal
and spatial key readings, (33), repeated here as (35):
(35) a. ∃E[∀y[kid(y) → ∃e∈E ∃π(e) [∀e’∈π(e) [∃X[two_monkeys(X) ∧
saw(y, X, e’) ]]]]]
b. |{X: two_monkeys(X) ∧ ∃y[kid(y)[saw(y, X, E)}|>1
The only difference between the event key readings and the participant key reading is
that for the temporal and spatial key readings the partitions are non-trivial. If the partition is
trivial, i.e π(e) = {e}, then all the monkey-sighting events by an individual kid are lumped
together. Thus, for every kid there will be just 2 monkeys that he or she saw. The plurality
condition then says that altogether there have to be more than two monkey pairs. The
interpretation in (35) then gives us all the three readings, the participant key reading, the
temporal key reading and the spatial key reading.
But does the postulation of the trivial partition in (34) bring about any unwanted
readings in the construction with the singular DP? I’ll show that it does not, in the next
section.
4.4 Lack of Participant key reading in the Singular DP construction
It was shown the RedNum construction with a singular DP has only the temporal key reading
and the spatial key reading, A participant key reading, with the distributive share, ‘2 2
monkeys’ trivially distributing over the ‘participant’, the subject DP ‘Ram’, is not possible
(36).
(36) ‘RAM SAW 2 2 MONKEYS’ = (9)
*Ram saw 2 monkeys. Participant key reading
Ram saw 2 monkeys in each time interval. Temporal key reading
Ram saw 2 monkeys in each location. Spatial key reading
According to the analysis that I have built up so far, the interpretation for this
sentence ‘Ram saw 2 2 monkeys’ will be (37).
Balusu – to appear in NELS 36 Proceedings 12
(37) a. ∃e ∃π(e) [∀e’∈π(e) ∃X[two_monkeys(X) ∧ saw(ram, X, e’)]]
b. |{X: two_monkeys(X) ∧ saw(ram, X, e)}|>1
There exists an event e and such a partition of e such that 2 monkeys were seen by
Ram in each of the cells of the partition. The number of monkeys seen by Ram in that e is
greater than 2. The plurality requirement in (37b) blocks the participant key reading, because
in the participant key reading there will be only two monkeys such that Ram saw them, and
the condition in (37b) requires that the number of monkeys seen by Ram be greater than 2.
This is a welcome result of the analysis. So in the cases where there is no plurality as a
‘participant’ (as in the plural and universal constructions), the trivial partition will not deliver
a participant key reading because of the plurality requirement. The plurality condition takes
care of this problem because it requires there to be more than one member of the distributive
share.
4.5 The readings with Plurals
Finally, I analyze the participant key reading in a RedNum construction with plurals, repeated
in (38) with the possible interpretations in (39).
(38) ‘[THE] KIDS SAW 2 2 MONKEYS’ = (11)
(39) a. The kids each saw 2 monkeys. Participant key reading
b. The kids saw 2 monkeys in each time interval. Temporal key reading
c. The kids saw 2 monkeys in each location. Spatial key reading
Along the lines of my analysis, the temporal key and spatial key readings come about
from the partitioning of the event into parts such that in each cell of the partition the kids saw
2 monkeys. The interpretation of these readings would be (40):
(40) a. ∃e ∃π(e) [∀e’∈π(e) ∃X[two_monkeys(X) ∧ saw(the kids, X, e’)]]
b. |{X: two_monkeys(X) ∧ saw (the kids, X, e)}|>1
There exists an e such that there is a partition of e in which 2 monkeys were seen by
the kids in each of the cells of the partition. The number of monkeys cumulatively seen by
the kids is greater than 2.
So how does the participant key reading arise? I propose that when predication with a
plural subject is interpreted as collective or cumulative, the construction with RedNum will
have the temporal key and the spatial key readings. When the predication is interpreted as
distributive, the participant key reading arises when the partition is construed as trivial. The
interpretation that we then get is (41):
(41) a. ∃E[∀y∈ the kids[∃e∈E[∀e’∈π(e) [∃X[two_monkeys(X) ∧
saw(y, X, e’) ]]]]]
Distributive Reduplication in Telugu 13
b. |{X: two_monkeys(X) ∧∃y[y∈the kids ∧ saw(y, X, E)}|>1
There is an event E such that for each of the kids there is an event e which is a part of
E such that there is a partition of e such that the kid saw 2 monkeys in every cell of e.
Altogether more than one monkey-pair was seen by the kids in E.
Note that there are four logical possibilities. We saw that the collective or cumulative
predication + non-trivial partition gives rise to the temporal and spatial key readings, the
distributive predication + trivial partition gives rise to the participant key reading. The
collective or cumulative predication + trivial partition combination will be ruled out by the
plurality requirement. The distributive predication + non-trivial partition combination gives
rise to readings that are parallel to the temporal and spatial key readings with the universal.
The participant key reading with plurals then arises because of two factors. (i) The
predicate ‘saw two-two monkeys’ is distributive with respect to the subject ‘the kids’, i.e. the
sentence has two D-operators: one coming from the predicate and one coming from RedNum.
(ii) There is just one cell in the partition of e for each kid – one that comprises all his/her
two-monkey sightings in e, i.e. the distribution associated with RedNum is redundant, the
partition is trivial.
The generalization about RedNum is now the following:
(42) The D-operator contributed by RedNum only takes events (event-aspects) as its
sorting key. A set of individuals is not accepted as its sorting key.
The so called “participant key” reading is also, according to this analysis, an eventkey
reading, but with a trivial partition of the event.
4.6 The main features of the proposal
The analysis that I have elaborated so far can be summed up as follows:
(43) (i) RedNum has a D-operator associated with it.
(ii) The D-operator takes an event or event-aspect as its sorting key
(iii) The event structure can be fine grained (non-trivial partition of e) or coarse
grained (trivial partition of e).
(iv) RedNum is associated with a plurality requirement.
5. Conclusion
In this paper I have proposed that RedNum is/has a D(istributivity)-operator which takes an
event or an event-aspect as its argument, and that it is associated with a plurality
requirement. This accounts for the temporal key readings and spatial key readings. I argue
that the “participant-key” reading is in fact also an event-key reading. The extra D-operator
comes from the predicate in the case of plural DP constructions or from the universal
quantifier in the construction with a universal QP. In these cases the action of the D-operator
Balusu – to appear in NELS 36 Proceedings 14
associated with RedNum is invisible because the event structure is coarse grained for these
cases, i.e. there is a trivial partition of the event associated with each participant. The analysis
is also able to correctly predict the multiple readings that are possible in constructions
involving more than one distributive numeral.
References
Choe, Jae-Woong. 1987. Anti-Quantifiers and a Theory of Distributivity. Doctoral
dissertation, University of Massachusetts, Amherst.
Doetjes, Jenny, and Martin Honcoop. 1997. The semantics of Event-related readings: a case
for pair quantification. In Ways of scope taking, ed. by Anna Szabolcsi, 263-310.
Dordrecht: Kluwer.
Landman, Fred. 1989. Groups. I. Linguistics and Philosophy 12(5):559–605.
Landman, Fred. 2000. Events and Plurality. Dordrecht: Kluwer.
Link, Godehard. 1998. Algebraic Semantics in Language and Philosophy. Vol. 74. Stanford:
CSLI Publications, Lecture Notes.
Zweig, Eytan. this volume. The implications of dependent plurals. In NELS 36, ed. by
Christopher Davis, Amy Rose Deal and Youri Zabbal. Amherst: University of
Massachusetts, GLSA.
Dept. of Linguistics
New York University
719 Broadway, 4th
floor
New York, NY 10003
rb964@nyu.edu