Derived collectives in Slavic: Decomposing
groups, swarms, and aggregates
Mojmír Dočekal & Marcin Wągiel
19.5.2017, University of Vienna
1 / 52
Collective nouns
Group nouns (Barker 1992) or bunch nouns (Schwarzschild
1996):
involve not only individuals, but also events and degrees
(1) a. group of objects
b. sequence of events
c. range of numbers
standard assumption: homogeneous category
plural denotation (Munn 1998, Elbourne 1999)
atomic denotation (Baker 1992, Schwarzschild 1996)
group-formation (Landman 1989, 2000)
2 / 52
Recent findings: distinct classes
Pearson (2011): committee nouns vs. collection nouns
Henderson (2017): group nouns vs. swarm nouns
Landman (2000):
collective body formation → swarms
collective action → groups/swarms
collective responsibility → groups
3 / 52
Groups vs. collections vs. swarms
Based on Pearson (2011) and Henderson (2017)
property groups collections swarms
plural pseudopartitives yes yes yes
plural agreement in BE/CE yes no no?
count Det heading partitives yes no no?
ILPs and always yes* no no?
only large constituent pluralities no yes? yes
spatial existence entailments no yes? yes
support spatial predicates no yes? yes
can expose a plurality yes no? mixed*
? our judgments
* unclear/dubious data
4 / 52
Plural pseudopartitives
Standard syntactic test (Barker 1992):
group noun + of-phrase with a plural complement →
group noun + of-phrase with a singular complement → *
(2) a. group of children/ * child
b. collection of shirts/ * shirt
c. set of chairs/ * chair
d. pair of shoes/ * shoe
e. herd of animals/ * animal
5 / 52
Problem:
expressions referring to game (Ritchie 2013)
(3) a. pride of lion
b. flock of pheasant
However:
English nouns referring to game can have a zero plural
(Corbett 2000: 68, Acquaviva 2008: 30)
(4) The elephant are downwind of us. (Allan 1976)
6 / 52
Nevertheless, the test is flawed anyway:
incorrect predictions wrt object (fake/neat) mass nouns
(5) a. group of offspring
b. collection of clothing
c. set of furniture
d. pair of footwear
e. herd of livestock
Hypothesis:
in pseudopartitive constructions group nouns require
nominals denoting pluralities
→ singular nouns do not refer to pluralities
7 / 52
Plural agreement and count Det heads
Possible with groups in BE/CE (Pearson 2011):
(6) a. The committee has been arguing all morning.
b. The committee have been arguing all morning.
(7) a. The committee is old. → ambiguous
b. The committee are old. → only distributive
(8) a. Three of the committee came to the meeting.
b. Several of the family objected to her marriage.
c. Many of the present cabinet will have to resign.
8 / 52
Impossible with collections:
(9) a. The deck of cards is on the table.
b. *The deck of cards are on the table.
(10) a. *Three of the bunch of flowers had died.
b. *Several of the deck of cards had gone missing.
c. *Many of the pile of dishes needed to be washed.
9 / 52
Absolutely impossible in Slavic:
(11) a. Komitet
committee
kłócił
argued.sg
się
refl
przez
through
całe
whole
rano.
morning
‘The committee has been arguing all morning.’
b. *Komitet
committee
kłocili
argued.pl
się
refl
przez
through
całe
whole
rano.
morning
Indended: ‘The committee have been arguing all
morning.’
(12) *Trzech
three
spośród
among
komitetu
committee
przyszło
came.sg
na
on
spotkanie.
meeting
Intended: ‘Three of the committee came to the
meeting.’
10 / 52
ILPs and always
Pearson (2011) → intensional semantics for group nouns
individual level predicates modified by always
(13) a. #John always has big feet. → object
b. Elephants always have big feet. → kind
(14) a. *That bunch of flowers is always tall.
b. The Pearson family always has big feet.
11 / 52
Problem:
it seems not all group nouns allow for that
(15) a. ??That group of students always has big feet.
b. ??That cast of actors always has big feet.
c. ??That cabinet always has big feet.
Hypothesis:
some groups involve temporal component → generations
of members
12 / 52
Derived collectives in Polish
Different classes:
suffix -ka → numerals
(16) dwa
two
→
→
dwójka
two.coll
‘two’ → ‘group of two’
suffix -stwo → animate nouns (human)
(17) rycerz
knight
→
→
rycerstwo
knight.coll
‘knight’ → ‘group of knights’
13 / 52
suffix -ostwo → animate nouns (social roles)
(18) wuj
uncle
→
→
wujostwo
uncle.coll
‘uncle’ → ‘uncle and his spouse’
suffix -ina → inanimate nouns (trees)
(19) brzoza
birch
→
→
brzezina
birch.coll
‘birch’ → ‘brich wood’
14 / 52
suffix -e → inanimate nouns
(20) kwiat
flower
→
→
kwiecie
flower.coll
‘flower’ → ‘mass of flowers’
there are more: -eria, -ela, -ba, -ota etc.
but they seem to pattern with -stwo
15 / 52
property -ka -stwo -ostwo -ina -e
derived from Numeral +ANIM N +ANIM N -ANIM N -ANIM
count yes no no yes no
cumulative reference no yes yes mixed yes
pseudopartitives yes no no no no
large const. pluralities no yes no yes yes
spatial existence ent. no no no yes yes
spatial predicates no no no yes no
16 / 52
Count/mass
Pluralization:
(21) a. dwójka
two.coll.sg
→
→
dwójki
two.coll.pl
b. rycerstwo
knight.col.sg
→
→
*rycerstwa
knight.coll.pl
c. wujostwo
uncle.coll.sg
→
→
*wujostwa
uncle.coll.pl
d. brzezina
birch.coll.sg
→
→
brzeziny
birch.coll.pl
e. kwiecie
flower.coll.sg
→
→
*kwiecia
flower.coll.pl
17 / 52
Numerals:
(22) a. dwie
two
dwójki
two.coll.pl
‘two groups of two’
b. *dwa
two
rycerstwa
knight.coll.pl
c. *dwa
two
wujostwa
uncle.coll.pl
d. dwie
two
brzeziny
birch.coll.pl
‘two birch groves’
e. *dwa
two
kwiecia
flower.coll.pl
18 / 52
Cumulative reference
(23) a. dwójka ⊕ dwójka = dwójka (‘two.coll’)
b. rycerstwo ⊕ rycerstwo = rycerstwo (‘knight.coll’)
c. wujostwo ⊕ wujostwo = wujostwo (‘uncle.coll’)
d. brzezina ⊕ brzezina = brzezina (‘birch.coll’) if not
connected
e. kwiat ⊕ kwiecie = kwiecie (‘flower.coll’)
19 / 52
Pseudopartitive constructions
The nominal root specifies constituents:
(24) a. dwójka
two.coll
chłopców
boys.gen
‘group of two boys’
b. *rycerstwo
knight.coll
jeźdźców
horsemen.gen
c. *wujostwo
uncle.coll
Austriaków
Austrians.gen
d. *brzezina
birch.coll
młodych
young
drzew
trees.gen
e. *kwiecie
flower.coll
niezpominajek
forget-me-nots.gen
20 / 52
Size of constituent pluralities
(25) a. dwójka (‘two.coll’) → cardinality = 2 (numeral
root → value)
b. rycerstwo (‘knight.coll’) → large cardinality
c. wujostwo (‘uncle.coll’) → cardinality = 2
(prototypically)
d. brzezina (‘birch.coll’) → large cardinality
e. kwiecie (‘flower.coll’) → large cardinality
21 / 52
Spatial existence entailments
If a group is dissolved, it ceases to exist (Henderson 2017):
groups survive spatial separation
swarms do not
(26) a. dwójka (‘two.coll’) → if separated
b. rycerstwo (‘knight.coll’) → if separated
c. wujostwo (‘uncle.coll’) → if separated
d. brzezina (birch.coll’) → if separated ×
e. kwiecie (‘flower.coll’) → if separated ×
22 / 52
Spatial predicates
(27) Context: particular individuals are arranged in such a
way to form a circle.
a. #Ta
this
dziesiątka
ten.coll
jest
is
okrągła.
circular
b. #To
this
rycerstwo
knight.coll
jest
is
okrągłe.
circular
c. #To
this
wujostwo
uncle.coll
jest
is
okrągłe.
circular
d. Ta
this
brzezina
birch.coll
jest
is
okrągła.
circular
‘This birch grove is circular.’
e. #To
this
kwiecie
flower.coll
jest
is
okrągłe.
circular
23 / 52
Different and same
interaction with derived collectives
intro on DS:
1. can be anaphoric to a referent introduced in previous
discourse – (28)
(28) Yesterday I bought a ticket for Kasabian in Vienna.
a. Today, Peter bought the same ticket.
b. Today, Peter bought a different ticket.
24 / 52
2. bound within a clause and express covariation (lack of
covariation)
(29) a. Each student bought the same ticket.
b. Each student bought a different ticket.
DS: anaphoric, different expresses covariation of tickets
with students
same – no covariation
Carlson term: sentence internal
25 / 52
Types of plural antecedents for DS
already Carlson (1987) noticed dependence of DS
acceptability on type of their plural antecedents:
(30) a. All the men are from different towns/?? a
different town.
b. Each man is from a different town/?? different
towns.
c. All the men are/each man is from the same town.
26 / 52
Cross-linguistic studies on DS:
Beck (2000), Tovena and van Peteghem 2002,
Matushansky 2008, Brasoveanu 2008
Beck’ (2000) data for German: sg. and pl. diferent in
English corresponds to two lexical items in German:
verschieden and andere
(31) a. Detmar
Detmar
und
and
Kordula
Kordula
wohnen
live
in
in
vershciedenen
different
Städten.
cities
b. Jedes
every
Mädchen
girl
hat
has
ein
a
anderes
different
Buch
book
gelesen.
read
27 / 52
Strategies for DS items:
Brasoveanu (2008): 11 languages study on DS lexical
items
1. sentence internal reading under distributive quantifiers
(plus discourse-anaphoric reading): andere, sg. different
2. only discourse-anaphoric reading: English other/another
3. sentence-internal reading with non-distributive quantifiers:
plural DP: verschidenen NPPL, different books
28 / 52
Dotlačil (2012): experimental confirmation of Brasoveaunu’s
claims – for Dutch
plus: strategy 3 acceptable with singular NP too:
(32) a. De
the
steden
towns
in
in
het
the
noorden
north
hebben
have
een
a
verschillende
verschillend
lengte-eenheid.
length-unit
b. Jan,
Jan,
Kees
Kees
en
and
Wim
Wim
hebben
have
een
a
verschillend
verschillend
schilderij
picture
gekozen.
chosen
29 / 52
Czech DS:
1. strategy 1 (distributive): jiný
2. strategy 2 (only discourse-anaphoric): ??? (ostatní?)
3. strategy 3 (non-distributive quantifiers): různý both in sg.
and pl.
introspection data:
(33) a. Každý
every
chlapec/???ti
boy/the
chlapci
boys
měl(i)
had
jiné
different
kolo.
bike
b.???Každý
every
chlapec/ti
body/the
chlapci
boys
měl(i)
had
různé(á)
different
kolo(a).
bike
30 / 52
plus Czech same which doesn’t distinguish between distributive
and non-distributive antecedents
general pattern for same cross-linguistically
(34) Každý
every
chlapec/ti
boy/the
chlapci
boys
měl(i)
had
stejné
same
kolo.
bike
31 / 52
Dotlačil (2012) formalization of the difference between
differentSG and different PL:
(35) a. differentPL = λPλx.#(x) ≥ 2(∀y, z <
x)[y, z ∈ AT ∧ y = z → distinct(y, z)]
b. differentSG = λPλx.Px ∧ ¬x ◦ y
c. same = λPλx.Px ∧ x = y
the observed dependence of sentence-internal reading for
differentSG: distributivity not part of different semantics
differentPL degraded with distributive antecedents . . . too
much distributivity in the sentence similar to:
(36) ???Each boy will each buy a ticket.
32 / 52
Czech reciprocals: the same strategies
1. R-reciprocals
reciprocals in Czech (introspection mostly)
R approach, does have distributivity built-in, with pure
distributive quantifiers the reciprocity is lost
(37) a. Petr a Marie se pozdravili.
Petr and Mary SE greeted (reciprocal ok)
b. Petr a Marie pozdravili jeden druhého.
Petr and Mary greeted each other (reciprocal ok)
(38) a. Každý policista se podezírá.
every policeman SE suspect (only reflexive)
b.???Každý policista podezírá jeden druhého.
every policeman suspect each other
33 / 52
R-reciprocals allow all types of reciprocity:
(39) a. Petr, Marie a Karel se znají.
Petr, Marie and Karel SE know (strong recipr.)
b. Petr, Marie a Karel bydlí 1 km od sebe.
Petr, Marie and Karel live 1 km from SE (weak)
c. Petr, Marie a Karel leželi na sobě.
Petr, Marie and Karel lied on SE (weakest)
34 / 52
R-type/differentPL: různý
(40) a. Ti žáci na to mají různý názor.
the students on that have different opinion
b.???Každý žák na to má různý názor.
every student on that has different opinion
35 / 52
2. DA approach (do not have distributivity built-in) –
distributive quantifiers help
(41) Každý policista podezírá ostatní policisty.
every policeman suspects other policemen
DA-reciprocals allow only the strongest reciprocity:
(42) a. Studenti, Petr, Marie a Karel znají ty ostatní
studenty.
Students, Petr, Marie and Karel know the other
students (recipr.)
b. Studenti, Petr, Marie a Karel bydlí 1 km od
ostatních studentů.
Students, Petr, Marie and Karel live 1 km from
other students (non-recipr.)
c. Petr, Marie a Karel leželi na ostatních
studentech.
Petr, Marie and Karel lied on other students 36 / 52
the same pattern for A-different: like DA-reciprocals are ok
with distributive quantifiers but their reciprocal reading
degrades with non-distributive NPs:
(43) a. Každý žák přečetl jinou knihu.
every student read other book (recipr. ok)
b. ??Ti žáci přečetli jinou knihu.
the students read other book (only non-recipr.)
37 / 52
same is acceptable both with collective and distributive
antecedents:
(44) a. Každý policista měl stejnou zbraň.
every policeman had same gun
b. Ti policisté měli stejnou zbraň.
the policemen had same gun
38 / 52
Derived mass nouns
Derived mass nouns derived by the suffix -í: list-í ’foliage’, dřív-í
’firewood’
Properties:
1. they are derived from -ANIM nouns
2. only singular: s list-í-mINST.SG vs. *s list-í-miINST.PL
3. not countable with cardinal numerals: *2 list-í, *2 dřív-í
and don’t allow switch to count (unlike underived mass
nouns: 2 vody, . . . )
But they can be counted with kind and sum numerals:
dvoj-í dříví, dvoj-e listí
4. the incompatibility with numerals comes from the
suffix/construction, not from the root: 2 list-y vs. *2 list-í
39 / 52
5. compatible with the singular universal quantifier všechno
’all’: všechno list-í, všechno dřív-í
6. obligatorily cumulative: list-í + list-í = list-í
7. obligatorily non-divisive (like plurals – the divisivity ends at
one unit): parts of list-í are list-í but not parts of list
8. topology plays a role: listí is plurality of connected objects,
. . . Grimm & Dočekal (2017)
40 / 52
Derived collective numerals
group nouns/numerals derived from cardinal numerals with
the suffix -ice: tr-oj-ice námořníků ’group-of-three sailors’
properties:
1. both singular and plural: s troj-icíINST.SG námořníků, s
troj-ice-miINST.PL námořníků
2. incompatible with the singular universal quantifier všechno
’all’: *všechna troj-ice námořníků
3. obligatorily non-cumulative: troj-ice + troj-ice = 2 troj-ice
4. obligatorily non-divisive: parts of troj-ice are not troj-ice
41 / 52
5. can be counted with cardinal numerals: dvě troj-ice
námořníků
6. can modify mostly +ANIM nouns (?metaphorical usages:
troj-ice úkolů ‘group-of-three tasks’, . . . )
7. topology is not involved: troj-ice detektivů se rozjela do tří
měst ‘group-of-three detectives departed into three towns’,
. . .
42 / 52
Derived collective/kind nouns
collective nouns derived from nouns of mostly professions
with the suffix -stvo: rytíř-stvo ‘knights/chivalry’,
duchovenstvo ‘clergy’
properties:
1. only singular: duchovenstvo, *duchovenstva
2. compatible with the singular universal quantifier všechno
‘all’: všechno rytířstvo
3. cumulative + divisive: rytířstvo + rytířstvo = rytířstvo,
parts of rytířstvo are still rytířstvo up to atomic knights?
4. countable to some extent with kind-level numerals: dvojí
duchovenstvo
5. mostly +HUMAN but some exceptions: loďstvo
‘marine/navy’
43 / 52
6. not topology but some sorts of institutionalization
7. somewhere between collective and kind:
ok with collective predicates like gather but un-countable
with cardinal numerals
ok with kind-level predicates like become extinct but
regular kinds (bear, . . . ) are more spatially dispersed
44 / 52
The experiment
Ways to design the experiment:
1. test same (introspection):
(45) a. ?Duchovenstvo mělo stejné čepice.
Clergy had same hats.
b. Duchovní měli stejné čepice.
Clergymen had same hats.
c. Každý duchovní měl stejnou čepici.
Every clergy had same hat.
45 / 52
2. Test differentSG (introspection):
(46) a.???Duchovenstvo mělo jinou čepici.
Clergy had different hat.
b.???Duchovní měli jinou čepici.
Clergymen had different hat.
c. Každý duchovní měl jinou čepici.
Every clergy had different hat.
46 / 52
3. Test differentPL (introspection):
(47) a. ?Duchovenstvo mělo různé čepice.
Clergy had different hats.
b. ?Duchovní měli různé čepice.
Clergymen had different hats.
c.???Každý duchovní měl různou čepici.
Every clergy had different hat.
we have chosen the strategy 2
47 / 52
(48) Hypothesis: collective nouns are decomposable to the
same extent as bare NPs.
the design: three groups of items
context always set to covariation scenario
each group has 9 items in three conditions
27 items in sum, plus 27 fillers
IBEX: randomization of conditions, reaction times, . . .
latin square design
each group represents one sub-type: derived mass nouns,
collective numerals, derived collective/kind nouns
48 / 52
Derived collective/kind nouns
(49) Kontext: Petr pozoroval poselstvo rytířů shromážděné
před hradem. Všiml si, že ani jeden rytíř nemá stejného
brnění jako ostatní. A řekl:
Context: Petr was observing messengerSTVO of knights
gathered in front of a castle. He noticed that no
knight has the same armour as others. And he said:
a. Poselstvo má jiné brnění.
MessengerSTVO has different armour.
b. Poslové mají jiné brnění.
MessengersPL have different armour.
c. Každý posel má jiné brnění.
Every messenger has different armour.
49 / 52
conditions in each item:
1. the target: derived collective/kind, derived mass, derived
coll. numeral
2. bare NP plural version of a)
3. universal quantified distributive quantifer + bare NP in
singular
expectations: c) should be reference level – nearly totally
acceptable
bare nouns and collective nouns should be statistically
significantly worse than distributive quantifier
real questions:
1. difference between a) and b)
2. difference between groups of items
50 / 52
Derived collective numerals
(50) Kontext: Petr pozoroval skupinu tří námořníků a všiml
si, že všichni mají odlišné oblečení. A řekl:
Context: Peter was observing a group of three sailors
and he noticed that all of them have different clothes.
And he said:
a. Trojice námořníků má jiné oblečení.
Group-of-three sailors has different clothes.
b. Námořníci mají jiné oblečení.
Sailors have different clothes.
c. Každý námořník má jiné oblečení.
Every sailor has different clothes.
51 / 52
Derived mass nouns
(51) Kontext: Petr pozoroval vzory cihloví na jedné vile a
všiml si, že ani jedna cihla neměla stejnou barvu jako
ostatní. A řekl:
Context: Petr looked at patterns of brick´I on a villa
and he noticed that no brick has the same color as
others. And he said:
a. Cihloví má jinou barvu.
Brick´I has different color.
b. Cihly mají jinou barvu.
Bricks have different color.
c. Každá cihla má jinou barvu.
Every brick has different color.
52 / 52