October 16 th 2015, Budapest Workshop on Linguistic and Cognitive - - PowerPoint PPT Presentation

October 16th 2015, Budapest Workshop on Linguistic and Cognitive aspects of Quantification

 The semantics and the acquisition of the

distributive marker po, which is notorious, (common to Slavic and has very broad distribution);

 our focus – po combining with numerals

in different argument positions (subject and object)

 What is the semantics of po?  How do adults and children understand

sentences with po combining with numerals (po-numerals)?

 a universal distributive quantifier (each)  a distributive-share marker (Choe 1987)  a distance-distributive marker (Zimmermann 2002)  a pluractional marker (Newman 1990, Lasersohn 1995)?

 Part 1: po is not a universal quantifier  Part 2: po quantifies over events  Part 3: experiment  Part 4: discussion

(i) Three kids are riding a horse.

Distributive Collective

 Most of the data come from the

languages with the universal distributive quantifier (e.g. each) (i) Three kids are riding a horse. Distributive: each of the three kids is riding a horse

PO IS NOT A UNIVERSAL QUANTIFIER

(1) Tri devojčice drže dva balona.

three girls hold two balloons

‘Three girls hold two balloons.’

collective: together distributive: separately/each

(2) Tri devojčice drže po dva balona.

three girls hold po two balloons

‘Three girls hold two balloons separately/each.’

‘Three girls hold two balloons at separate

locations/time intervals.’

# collective: together distributive: separately/each distributive: at separate locations/time intervals

participant distributive ‘Three girls hold two balloons separately/each.’ > distribution is over event participants (girls) (cf. Cable 2014)

event-distributive ‘Three girls hold two balloons at separate locations/time intervals.’ > distribution is over locations/times

(2) [Devojčice]key drže [po dva balona]share

girls hold po two balloon

(3) [Each girl]key holds [two balloons]share

• Share is what is distributed
• Key is distributed over

(2) [Devojčice]key drže [po dva balona]share

girl hold po two balloon

 po attaches to the share (what is distributed)

(3) [Each girl]key hold [two balloons]share

 each (determiner-like) attaches to the key (the NP over

which the distribution takes place)

[Each girl] [holds two balloons]. [Devojčice] [drže po dva balona].

girls hold po two balloons

x [x is a girl] [(y) y is ‘2 balloons’  x holds y]

(3) Each girl holds two balloons.

 Exhaustivity (goup of girls needs to be exhausted)  Atomicity

(the group of girls needs to be partitioned into

atoms)

(3) Each girl holds two balloons.

 Context A: Mary and Jane hold 2

balloons together and Rose holds 2 balloons ✗

 Context B: Mary and Jane hold 2

balloons each while Rose is holding 2 kites ✗

(2) Devojčice drže po dva balona.

girls hold po two balloon

 no exhaustivity requirement

(the group of girls needs not be exhausted)

 no atomicity requirement

(the group of girls needs not be partitioned in atoms)

(2) Devojčice drže po dva balona

girls hold po two balloon

 Context A: Mary and Jane hold 2

balloons together and Rose holds 2 balloons ✓

 Context B: Mary and Jane hold 2

balloons each while Rose is holding 2 kites ✓

Differences

 Spatial/temporal distribution  ‘Non-exhaustive’ context  ‘Non-atomic’ context  po ≠ each, UQ or DD (floated)

 one way of salvaging the key-share

view: exhaustivity requirement is over events (the key is always an event - spatiotemporal argument)

 hard to disprove or test

G1: No atomic partitioning requirement

 Participants in the described events do not

have to be atomically partitioned (they can also be partitioned into groups or not partitioned at all). G2: No exhaustivity requirement

 The group of participants in the event

needs not be exhausted (‘exhaustively distributed over’). This holds even in the case where the distribution is over atomic participants.

 po is not a universal distributive quantifier  po attaches to what is distributed

(distshare)

PO QUANTIFIES OVER EVENTS

Claim

 po systematically involves distribution

• ver spaces/times

 participant-distribution follows from

event-distribution

(1) Tri devojčice drže po dva balona (2) Three girls each hold two balloons. ‘Each of the three girls holds two balloons.’

• participant-distributive (exhaustive and

atomic)

(1) Tri devojčice drže po dva balona (2) Three girls each hold two balloons. ‘Some (but not necessary all) girls holds two balloons in groups or individually.’

• participant-distributive (not exhaustive

and not atomic)

(1) Tri devojčice drže po dva balona. (2) Three girls each hold two balloons. ‘Three girls hold two balloons at separate locations or/and time intervals.’

• event-distributive

 po semantically combines first with a

numeral (n) and then with an NP and

 there must be at least two events that

involve an NP of the cardinality n (cumulation of events involving nNP, ex. po 2 balloons)

[[po]] = n. Q<e,t>. P<e, t>. e. e  *nQ & e  nQ & x. Q(x) = 1 & P(x)(e) = 1

Tri devojčice drže po dva balona three girls hold po two balloons

• A plurality of events – constructed out of

single events involving 2 balloons

hold-event agent theme e1 Mary b1+b2 e2 Jane b3+b4 e3 Rose b5+b6 *hold-event_ agent-sum theme-sum e1+e2 Mary+Jane b1+b2+b3+b4 e1+e2+e3 Mary+Jane+Rose b1+b2+ b3+b4 +b5+b6 …

• at least two events involving 2 balloons

hold-event agent theme e1 Mary+Jane+Rose b1+b2 e2 Mary+Jane+Rose b3+b4 e3 Mary+Jane+Rose b5+b6 *hold-event_ agent-sum theme-sum e1+e2+e3 Mary+Jane+Rose b1+b2+ b3+b4 +b5+b6 …

• at least two events involving 2 balloons

 po is not a universal distributive quantifier

like each, since it does not require atomic partitioning of the group of participants or exhaustive distribution over participants

 po quantifies over spatiotemporal units

(events)

 quantification over events derives so-called

quantification over participants

 the participants can but need not be

‘atomic’ or ‘exhausted’

ACQUISITION

Do children know that po yields:

 event-distributive readings (distribution

• ver space/time)

 participant-distributive readings (events

involving atomic participants) and

 syntactically attaches to what is

distributed (distributive share, cf. Choe 1987)?

(1) po-object sentences Tri devojčice drže po dva balona. three girls hold po two balloon ‘Three girls are (each) / separately holding two balloons.’ (2) po-subject sentences Po tri devojčice drže dva balona. po three girls hold two balloon ‘Three girls are holding each of two balloons.’ (3) double-po sentences Po tri devojčice drže po dva balona. po three girls hold po two balloon ‘Three girls are holding two balloons at separate locations.’

Subject-distributive (SD) (1) Tri devojčice drže po dva balona. three girls hold po two balloons ‘Three girls are (each) / separately holding two balloons.’

Object-distributive (OD) Po tri devojčice drže dva balona. po three girls hold two balloons ‘Three girls are holding each of two balloons.’

Event-distributive (ED) (3) Po tri devojčice drže po dva balona. po three girls hold po two balloons ‘Three girls are holding two balloons at separate locations.’

SD OD ED po-object

YES NO YES

po-subject

NO YES YES

double-po

NO NO YES

 Children: N=19, between ages 6;8 and

7;6 (MA= 6.5, SD = 0.5,10 girls)

 Adults: N=17, f=12, MA= 35.5, sd=8.8

 54 test items  18 distracters  block design (24 items) balanced across

subjects

 TVJT

 H1: All po-sentences allow event-

distributive readings.

 H2: The position of po determines the

type of a participant-distributive reading

• f the sentence (subject-distributive or
• bject-distributive).

10 20 30 40 50 60 70 80 90 100 po-object po-subject double-po SD OD ED

 The syntactic position of po determines

the participant-distributive reading of its sentence (supporting H2)

 Object-distributive readings (OD) were

accepted significantly less than all other readings (unpredicted)

the information structure hypothesis:

 Left-most peripheral positions are

suitable topics (Godjevac 2003)

 Nonspecific NPs are not suitable

candidates for topics

 po-phrases are nonspecific (Choe 1987)

Well formed: (1)a. Tri devojčice drže po dva balona.

three girls hold po two balloons

• b. [TopP [tri devojčice]1 [IP t1 drže po dva balona] ]

Ill-formed (2) Po tri devojčice drže dva balona.

po three girls hold two balloons

 po-phrase is non-specific (unsuitable topic) and occupies the

left-most position (topic position)

 Adults do not like po-initial sentences

and not OD contexts

prediction:

[TopP [dva balona]1 [IP po tri devojčice drže t1] ] two balloons po three girls hold

10 20 30 40 50 60 70 80 90 100 po-object po-subject double-po SD OD ED

G1:Children accept both participant- distributive and event-distributive readings G2:Children are not sensitive to the locus of po G3:Children do not like distributing subjects alone (without objects)

 G1 and G2 – children do not know the

meaning of po

 G3: Children do not like distributing

subjects alone (without objects)

 Why?

10 20 30 40 50 60 70 80 90 100 po-object po-subject double-po Children Adults

20.2% vs. 47.1%, 2(1) = 16.4, p < 0.001

 sentences with two numerals involve

scope relations between two nNPs

 the inverse scope construal (object-

distributive) is disprefered to surface scope (subject-distributive)

But

 our analysis of distributivity in Serbian

does not rely on scope - the scope of two nNPs is not relevant for the resulting (participant) distributive reading

 Could it be that scope is relevant for

children?

 NO

 The truth-conditions of double-po

sentences are very problematic for scopal analysis, but follow from event quantification

 Children accept double-po sentences at

high rates

(3) Po three girls hold po two balloons

• true in the scenario: every day, three girls

together hold two balloons > follows from event quantification analysis but not from scopal analysis

 thematic prominence of arguments,

may be more important for children than scope relations (Lee 2003, É. Kiss et al. 2012)

Hierarchy of thematic roles in children (Lee 2003; cf. É. Kiss et al. 2012) ACTOR > LOCATION > PATIENT

Children at age 6:

 know that po yields event-distributive

and participant-distributive readings;

 have no problems with double-po

sentences and ED reading Claim: po quantifies over events and participants-distribution should be reduced to event-distribution

Acquisition:

 How to further experimentaly investigate

this claim

 Temporal distribution with po

Theory:

 Other instances of po

 locative preposition (spatiotemporal

meaning) Attaches also to:

• numerals ‘po tri’ (po three)
• weak quantifiers ‘po mnogo’ (po many)
• bare singular nouns ‘po brod’ (po boat)
• verbs, adverbs, adjectives

 systematically introduces spatiotemporal

separation/grouping/distribution (among other meanings)

THANK YOU! Thanks to Ana Bosnić for data collection