Non-grammaticalized number entails an exclusive interpretation of - - PowerPoint PPT Presentation

Non-grammaticalized number entails an exclusive interpretation of plural morphology

Adam Liter adam.liter@gmail.com

• A. Tess Huelskamp

huelska1@msu.edu Christopher C. Heffner heffner@umd.edu Cristina Schmitt schmit12@msu.edu Workshop on Linguistic and Cognitive Aspects of Quantification 2015 October 17, 2015

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 1 / 36

Introduction

The question

Why do certain quantificational elements have the interpretations that they have?

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 2 / 36

Introduction

Goal of the paper

Examine to what extent inclusive/exclusive interpretations of the plural are the result of properties of the input. To do this, we use two versions of an artificial language that we will call Lablish I (Liter et al. 2014) and Lablish II.

◮ Lablish I, II, and English are qualitatively different:

English has grammaticalized number; OBLIGATORY singular and plural distinctions.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 3 / 36

Introduction

Goal of the paper

Examine to what extent inclusive/exclusive interpretations of the plural are the result of properties of the input. To do this, we use two versions of an artificial language that we will call Lablish I (Liter et al. 2014) and Lablish II.

◮ Lablish I, II, and English are qualitatively different:

English has grammaticalized number; OBLIGATORY singular and plural distinctions. Lablish I and II do NOT have grammaticalized number; OPTIONAL singular and plural distinctions, much like Korean, Japanese and Mandarin.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 3 / 36

Introduction

Goal of the paper

Examine to what extent inclusive/exclusive interpretations of the plural are the result of properties of the input. To do this, we use two versions of an artificial language that we will call Lablish I (Liter et al. 2014) and Lablish II.

◮ Lablish I, II, and English are qualitatively different:

English has grammaticalized number; OBLIGATORY singular and plural distinctions. Lablish I and II do NOT have grammaticalized number; OPTIONAL singular and plural distinctions, much like Korean, Japanese and Mandarin.

◮ Lablish I and II are quantitatively different: Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 3 / 36

Introduction

Goal of the paper

Examine to what extent inclusive/exclusive interpretations of the plural are the result of properties of the input. To do this, we use two versions of an artificial language that we will call Lablish I (Liter et al. 2014) and Lablish II.

◮ Lablish I, II, and English are qualitatively different:

English has grammaticalized number; OBLIGATORY singular and plural distinctions. Lablish I and II do NOT have grammaticalized number; OPTIONAL singular and plural distinctions, much like Korean, Japanese and Mandarin.

◮ Lablish I and II are quantitatively different:

Lablish I: 50% number-neutral bare nouns, 25% singular NPs and 25% plural NPs.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 3 / 36

Introduction

Goal of the paper

Examine to what extent inclusive/exclusive interpretations of the plural are the result of properties of the input. To do this, we use two versions of an artificial language that we will call Lablish I (Liter et al. 2014) and Lablish II.

◮ Lablish I, II, and English are qualitatively different:

English has grammaticalized number; OBLIGATORY singular and plural distinctions. Lablish I and II do NOT have grammaticalized number; OPTIONAL singular and plural distinctions, much like Korean, Japanese and Mandarin.

◮ Lablish I and II are quantitatively different:

Lablish I: 50% number-neutral bare nouns, 25% singular NPs and 25% plural NPs. Lablish II: 12.5% number-neutral bare nouns, 43.75% singular NPs and 43.75% plural NPs.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 3 / 36

Introduction

About artificial language experiments

What is bad about them:

◮ Artificial languages are artificial. ◮ The experiments are always L2 experiments.

What is good about them:

◮ Artificial languages have controlled/cleaned up input. ◮ If we assume that the learner’s deviations from the input indicate learners’

• wn biases about natural language structure, then artificial language

experiments may help us understand typological properties of natural languages as being imposed by the learning mechanisms.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 4 / 36

Introduction

Artificial language paradigms and what we have learned from them

Type I: Competition between typologically preferred and dispreferred properties/structures.

◮ Measurement: Mastery of the preferred/dispreferred properties/structures. ◮ Finding: Common/preferred properties/structures are learned faster and

more easily.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 5 / 36

Introduction

Artificial language paradigms and what we have learned from them

Type I: Competition between typologically preferred and dispreferred properties/structures.

◮ Measurement: Mastery of the preferred/dispreferred properties/structures. ◮ Finding: Common/preferred properties/structures are learned faster and

more easily.

Type II: Probabilistic marking of a particular syntactic property (case).

◮ Measurement: Accuracy in matching frequencies in the input. ◮ Finding: Subjects tend to reorganize the system following some universal

tendency (DOM) (Fedzechkina et al. 2012).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 5 / 36

Introduction

Artificial language paradigms and what we have learned from them

Type I: Competition between typologically preferred and dispreferred properties/structures.

◮ Measurement: Mastery of the preferred/dispreferred properties/structures. ◮ Finding: Common/preferred properties/structures are learned faster and

more easily.

Type II: Probabilistic marking of a particular syntactic property (case).

◮ Measurement: Accuracy in matching frequencies in the input. ◮ Finding: Subjects tend to reorganize the system following some universal

tendency (DOM) (Fedzechkina et al. 2012).

Type III: Optional use of forms that are in semantic competition and interpretational inferences (number).

◮ Measurement: Knowledge of optionality (the plural/singular distinction is

marked optionally) and possible inferences in a novel context for the learner (downward entailing context).

◮ Finding: The input influences how plural is interpreted in downward

entailing contexts.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 5 / 36

Roadmap

Roadmap

1 About plurals 2 Explaining crosslinguistic differences 3 Study I and Study II

Hypotheses Methods Results

4 Discussion

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 6 / 36

About plurals

About plurals: inclusive vs. exclusive interpretations

The plural in English doesn’t always mean “more than one”. (1) John didn’t see friends last night.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 7 / 36

About plurals

About plurals: inclusive vs. exclusive interpretations

The plural in English doesn’t always mean “more than one”. (1) John didn’t see friends last night. English allows an inclusive interpretation of the plural in downward entailing contexts.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 7 / 36

About plurals

About plurals: inclusive vs. exclusive interpretations

The plural in English doesn’t always mean “more than one”. (1) John didn’t see friends last night. English allows an inclusive interpretation of the plural in downward entailing contexts. Korean plural does seem to always mean “more than one”. (2) con-un John-top eceyspam last.night chinkwu-tul-ul friend-pl-acc manna-ci meet-nmlz anha-ss-ta. neg-pst-decl ‘John didn’t see friends last night.’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 7 / 36

About plurals

About plurals: inclusive vs. exclusive interpretations

The plural in English doesn’t always mean “more than one”. (1) John didn’t see friends last night. English allows an inclusive interpretation of the plural in downward entailing contexts. Korean plural does seem to always mean “more than one”. (2) con-un John-top eceyspam last.night chinkwu-tul-ul friend-pl-acc manna-ci meet-nmlz anha-ss-ta. neg-pst-decl ‘John didn’t see friends last night.’ Korean has an exclusive interpretation of the plural in all contexts.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 7 / 36

About plurals

An account of these facts

Sauerland et al. (2005) argues that the English plural is meaningless (“semantically unmarked”).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 8 / 36

About plurals

An account of these facts

Sauerland et al. (2005) argues that the English plural is meaningless (“semantically unmarked”). The exclusive interpretation in many English contexts is derived by an implicated presupposition (see also Sauerland 2008a).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 8 / 36

About plurals

An account of these facts

Sauerland et al. (2005) argues that the English plural is meaningless (“semantically unmarked”). The exclusive interpretation in many English contexts is derived by an implicated presupposition (see also Sauerland 2008a). The implicated presupposition is calculated in accordance with Maximize Presupposition, as proposed by Heim (1991). (3) Maximize Presupposition Make your contribution presuppose as much as possible.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 8 / 36

About plurals

An account of these facts

Sauerland et al. (2005) argues that the English plural is meaningless (“semantically unmarked”). The exclusive interpretation in many English contexts is derived by an implicated presupposition (see also Sauerland 2008a). The implicated presupposition is calculated in accordance with Maximize Presupposition, as proposed by Heim (1991). (3) Maximize Presupposition Make your contribution presuppose as much as possible. The implicated presupposition is cancelled in downward entailing contexts, allowing English to have an inclusive interpretation of the plural in such contexts.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 8 / 36

About plurals

An account of these facts

Sauerland et al. (2005) argues that the English plural is meaningless (“semantically unmarked”). The exclusive interpretation in many English contexts is derived by an implicated presupposition (see also Sauerland 2008a). The implicated presupposition is calculated in accordance with Maximize Presupposition, as proposed by Heim (1991). (3) Maximize Presupposition Make your contribution presuppose as much as possible. The implicated presupposition is cancelled in downward entailing contexts, allowing English to have an inclusive interpretation of the plural in such contexts. The Korean plural cannot be semantically unmarked; it must mean “more than one”.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 8 / 36

Explaining crosslinguistic differences

Why do pluralizers differ crosslinguistically?

Are the different interpretations the result of a mere historical accident?

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 9 / 36

Explaining crosslinguistic differences

Why do pluralizers differ crosslinguistically?

Are the different interpretations the result of a mere historical accident? Or are these different interpretations something that follow from properties of the grammar?

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 9 / 36

Explaining crosslinguistic differences

Why do pluralizers differ crosslinguistically?

Are the different interpretations the result of a mere historical accident? Or are these different interpretations something that follow from properties of the grammar? If so, what is the relevant property?

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 9 / 36

Explaining crosslinguistic differences

Why the exclusive interpretation?

Two possibilites:

(i) Non-grammaticalized number does not license semantic unmarkedness. (ii) Semantic competition blocks semantic unmarkedness (ban on synonyms).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 10 / 36

Explaining crosslinguistic differences

(Non-)grammaticalized number and semantic unmarkedness

Kwon and Zribi-Hertz (2004):

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 11 / 36

Explaining crosslinguistic differences

(Non-)grammaticalized number and semantic unmarkedness

Kwon and Zribi-Hertz (2004):

◮ English has an inflectional functional number head with both positive and

negative values (e.g., +pl and −pl or +sg and −sg).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 11 / 36

Explaining crosslinguistic differences

(Non-)grammaticalized number and semantic unmarkedness

Kwon and Zribi-Hertz (2004):

◮ English has an inflectional functional number head with both positive and

negative values (e.g., +pl and −pl or +sg and −sg).

◮ The Korean plural is a lexical non-functional plural marker with no

corresponding “negative” value.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 11 / 36

Explaining crosslinguistic differences

(Non-)grammaticalized number and semantic unmarkedness

Kwon and Zribi-Hertz (2004):

◮ English has an inflectional functional number head with both positive and

negative values (e.g., +pl and −pl or +sg and −sg).

◮ The Korean plural is a lexical non-functional plural marker with no

corresponding “negative” value.

In a φ-feature paradigm, one of the values is semantically unmarked (Sauerland 2008b).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 11 / 36

Explaining crosslinguistic differences

(Non-)grammaticalized number and semantic unmarkedness

Kwon and Zribi-Hertz (2004):

◮ English has an inflectional functional number head with both positive and

negative values (e.g., +pl and −pl or +sg and −sg).

◮ The Korean plural is a lexical non-functional plural marker with no

corresponding “negative” value.

In a φ-feature paradigm, one of the values is semantically unmarked (Sauerland 2008b). If correct, then:

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 11 / 36

Explaining crosslinguistic differences

(Non-)grammaticalized number and semantic unmarkedness

Kwon and Zribi-Hertz (2004):

◮ English has an inflectional functional number head with both positive and

negative values (e.g., +pl and −pl or +sg and −sg).

◮ The Korean plural is a lexical non-functional plural marker with no

corresponding “negative” value.

In a φ-feature paradigm, one of the values is semantically unmarked (Sauerland 2008b). If correct, then:

◮ grammaticalization licenses semantic unmarkedness; Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 11 / 36

Explaining crosslinguistic differences

(Non-)grammaticalized number and semantic unmarkedness

Kwon and Zribi-Hertz (2004):

◮ English has an inflectional functional number head with both positive and

negative values (e.g., +pl and −pl or +sg and −sg).

◮ The Korean plural is a lexical non-functional plural marker with no

corresponding “negative” value.

In a φ-feature paradigm, one of the values is semantically unmarked (Sauerland 2008b). If correct, then:

◮ grammaticalization licenses semantic unmarkedness; ◮ and non-grammaticalization does not. Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 11 / 36

Explaining crosslinguistic differences

Semantic competition and semantic unmarkedness

Another difference:

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 12 / 36

Explaining crosslinguistic differences

Semantic competition and semantic unmarkedness

Another difference:

◮ English has obligatorily singular-marked or plural-marked NPs; Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 12 / 36

Explaining crosslinguistic differences

Semantic competition and semantic unmarkedness

Another difference:

◮ English has obligatorily singular-marked or plural-marked NPs; ◮ whereas number-neutral bare nominals are most frequent in Korean and

singular- and plural-marked NPs are infrequent.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 12 / 36

Explaining crosslinguistic differences

Semantic competition and semantic unmarkedness

Another difference:

◮ English has obligatorily singular-marked or plural-marked NPs; ◮ whereas number-neutral bare nominals are most frequent in Korean and

singular- and plural-marked NPs are infrequent.

The denotation of the bare NP in Korean contains the singular- and plural-marked NPs. It is unspecified for number.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 12 / 36

Explaining crosslinguistic differences

Semantic competition and semantic unmarkedness

Another difference:

◮ English has obligatorily singular-marked or plural-marked NPs; ◮ whereas number-neutral bare nominals are most frequent in Korean and

singular- and plural-marked NPs are infrequent.

The denotation of the bare NP in Korean contains the singular- and plural-marked NPs. It is unspecified for number. Given how frequent it is, this blocks the learner from assigning the same (lack of) meaning to the plural in Korean (cf. Brazilian Portuguese).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 12 / 36

Study I and Study II Hypotheses

Predictions

In a replication of Liter et al. (2014) with decreased frequency of number-neutral bare NPs, participants should:

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 13 / 36

Study I and Study II Hypotheses

Predictions

In a replication of Liter et al. (2014) with decreased frequency of number-neutral bare NPs, participants should:

◮ assign an exclusive interpretation to the plural under the scope of negation

(non-grammaticalized number hypothesis).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 13 / 36

Study I and Study II Hypotheses

Predictions

In a replication of Liter et al. (2014) with decreased frequency of number-neutral bare NPs, participants should:

◮ assign an exclusive interpretation to the plural under the scope of negation

(non-grammaticalized number hypothesis).

◮ assign an inclusive interpretation to the plural under the scope of negation

(semantic competition hypothesis).

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 13 / 36

Study I and Study II Methods

Methods

Modulate frequencies across studies:

Study I Study II Heard Saw % % NP-∅ Singleton 50% 12.5% NP-∅ Plurality NP-paya Singleton 25% 43.75% NP-koho Plurality 25% 43.75%

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 14 / 36

Study I and Study II Methods

Methods

Modulate frequencies across studies:

Study I Study II Heard Saw % % NP-∅ Singleton 50% 12.5% NP-∅ Plurality NP-paya Singleton 25% 43.75% NP-koho Plurality 25% 43.75%

The language:

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 14 / 36

Study I and Study II Methods

Methods

Modulate frequencies across studies:

Study I Study II Heard Saw % % NP-∅ Singleton 50% 12.5% NP-∅ Plurality NP-paya Singleton 25% 43.75% NP-koho Plurality 25% 43.75%

The language:

◮ 16 nouns Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 14 / 36

Study I and Study II Methods

Methods

Modulate frequencies across studies:

Study I Study II Heard Saw % % NP-∅ Singleton 50% 12.5% NP-∅ Plurality NP-paya Singleton 25% 43.75% NP-koho Plurality 25% 43.75%

The language:

◮ 16 nouns ◮ 2 transitive verbs; 2 intransitive verbs Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 14 / 36

Study I and Study II Methods

Methods

Modulate frequencies across studies:

Study I Study II Heard Saw % % NP-∅ Singleton 50% 12.5% NP-∅ Plurality NP-paya Singleton 25% 43.75% NP-koho Plurality 25% 43.75%

The language:

◮ 16 nouns ◮ 2 transitive verbs; 2 intransitive verbs ◮ 2 number morphemes (paya ‘sg’; koho ‘pl’) Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 14 / 36

Study I and Study II Methods

Methods

Modulate frequencies across studies:

Study I Study II Heard Saw % % NP-∅ Singleton 50% 12.5% NP-∅ Plurality NP-paya Singleton 25% 43.75% NP-koho Plurality 25% 43.75%

The language:

◮ 16 nouns ◮ 2 transitive verbs; 2 intransitive verbs ◮ 2 number morphemes (paya ‘sg’; koho ‘pl’) ◮ 1 negation particle Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 14 / 36

Study I and Study II Methods

Methods

Modulate frequencies across studies:

Study I Study II Heard Saw % % NP-∅ Singleton 50% 12.5% NP-∅ Plurality NP-paya Singleton 25% 43.75% NP-koho Plurality 25% 43.75%

The language:

◮ 16 nouns ◮ 2 transitive verbs; 2 intransitive verbs ◮ 2 number morphemes (paya ‘sg’; koho ‘pl’) ◮ 1 negation particle ◮ Word order: (neg)-VSO Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 14 / 36

Study I and Study II Methods

Training and testing schedule

There were 10 sessions, 2 of which were exclusively testing sessions.

Task 1 2 3 4 5 6 7 8 9 10 Training

• General grammar
• Vocab & number morphology use
• Truth value judgment
• Verb & object negation
• Number negation
• Free response
• Liter et al.

Non-grammaticalized number & plural October 17 – LCQ 2015 15 / 36

Study I and Study II Methods

Basic training

Participants would hear something like in (4) and see the accompanying event on the screen. (4) beguku circle hatepi cow paya sg penana snake

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 16 / 36

Study I and Study II Methods

Verb negation training

‘jiggle dog-pl’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 17 / 36

Study I and Study II Methods

Verb negation training

‘jiggle dog-pl’ ‘bounce dog-pl’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 17 / 36

Study I and Study II Methods

Verb negation training

‘jiggle dog-pl’ ‘bounce dog-pl’ ‘neg jiggle dog-pl’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 17 / 36

Study I and Study II Methods

Verb negation training

‘jiggle dog-pl’ ‘bounce dog-pl’ ‘neg jiggle dog-pl’ ‘bounce dog-pl’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 17 / 36

Study I and Study II Methods

Object negation training

‘circle lizard rabbit’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 18 / 36

Study I and Study II Methods

Object negation training

‘circle lizard rabbit’ ‘circle lizard girl’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 18 / 36

Study I and Study II Methods

Object negation training

‘circle lizard rabbit’ ‘circle lizard girl’ ‘neg circle lizard girl’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 18 / 36

Study I and Study II Methods

Object negation training

‘circle lizard rabbit’ ‘circle lizard girl’ ‘neg circle lizard girl’ ‘circle lizard elephant’

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 18 / 36

Study I and Study II Methods

Participants

Study I:

◮ 20 adult native English speakers participated. ◮ Only 17 completed the experiment.

Study II:

◮ 16 adult native English speakers participated. ◮ 16 completed the experiment. Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 19 / 36

Study I and Study II Results

Grammar task results

0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II 2 3 4 5 6 7 8 10

Session Accuracy

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 20 / 36

Study I and Study II Results

Free response task

BareForSG PayaForSG BareForPL KohoForPL 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II 7 10 7 10 7 10 7 10

Session Frequency of use

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 21 / 36

Study I and Study II Results

Vocabulary and number morphology use task

Participants hear ‘cow-sg’ and have to choose either the single cow, the two cows, both sets of cows, or neither set of cows.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 22 / 36

Study I and Study II Results

Basic noun recall

0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II 2 3 4 5 6 7 8 9 10

Session Accuracy

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 23 / 36

Study I and Study II Results

Interpretation of NP types (NP-pl, NP-∅, and NP-sg)

NP-pl

0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II 2 3 4 5 6 7 8 9 10

Session Frequency of response

Response type Both PL & SG PL SG

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 24 / 36

Study I and Study II Results

Interpretation of NP types (NP-pl, NP-∅, and NP-sg)

NP-sg

0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II 2 3 4 5 6 7 8 9 10

Session Frequency of response

Response type Both PL & SG PL SG

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 24 / 36

Study I and Study II Results

Interpretation of NP types (NP-pl, NP-∅, and NP-sg)

NP-∅

0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II 2 3 4 5 6 7 8 9 10

Session Frequency of response

Response type Both PL & SG PL SG

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 24 / 36

Study I and Study II Results

NP-∅ truth value judgment task

Three trial types:

◮ FalseTrials ◮ TrueSingularTrials ◮ TruePluralTrials

FalseTrials TrueSingularTrials TruePluralTrials 0.00 0.25 0.50 0.75 1.00

Frequency of response

false true

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 25 / 36

Study I and Study II Results

Verb negation

TrueTrials FalseTrials 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II

Frequency of response

false true

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 26 / 36

Study I and Study II Results

Object negation

TrueTrials FalseTrials 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II

Frequency of response

false true

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 27 / 36

Study I and Study II Results

Number negation

Participants considered in the number negation task:

◮ Study I: 12 participants ◮ Study II: 5 participants Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 28 / 36

Study I and Study II Results

Number negation trial types

payaXsg te neg beguku circle tomidi ladybug keka duck paya sg ‘A ladybug isn’t/aren’t circling a duck’ Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 29 / 36

Study I and Study II Results

Number negation trial types

payaXsg te neg beguku circle tomidi ladybug keka duck paya sg ‘A ladybug isn’t/aren’t circling a duck’ payaXpl te neg disi cover mamuki zebra bo boy paya sg ‘A zebra isn’t/aren’t covering a boy’ Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 29 / 36

Study I and Study II Results

Number negation trial types

payaXsg te neg beguku circle tomidi ladybug keka duck paya sg ‘A ladybug isn’t/aren’t circling a duck’ payaXpl te neg disi cover mamuki zebra bo boy paya sg ‘A zebra isn’t/aren’t covering a boy’ kohoXsg te neg disi cover mibohe lion daka girl koho pl ‘A lion isn’t/aren’t covering girls’ Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 29 / 36

Study I and Study II Results

Number negation trial types

payaXsg te neg beguku circle tomidi ladybug keka duck paya sg ‘A ladybug isn’t/aren’t circling a duck’ payaXpl te neg disi cover mamuki zebra bo boy paya sg ‘A zebra isn’t/aren’t covering a boy’ kohoXsg te neg disi cover mibohe lion daka girl koho pl ‘A lion isn’t/aren’t covering girls’ kohoXpl te neg beguku circle sopoka rabbit nupu mouse koho pl ‘A rabbit isn’t/aren’t circling mice’ Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 29 / 36

Study I and Study II Results

Number negation results

Participants more likely than chance to say true to kohoXsg trials:

◮ Study I:

χ2(1) = 50.776, p < 0.001

◮ Study II:

χ2(1) = 20.833, p < 0.001

payaXPL payaXSG kohoXPL kohoXSG 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II

Frequency of response

false true

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 30 / 36

Discussion

Discussion

Participants overwhelmingly said kohoXsg trials were true.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 31 / 36

Discussion

Discussion

Participants overwhelmingly said kohoXsg trials were true. This suggests they are assigning an exclusive interpretation to the plural, like in Korean and unlike in English.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 31 / 36

Discussion

Discussion

Participants overwhelmingly said kohoXsg trials were true. This suggests they are assigning an exclusive interpretation to the plural, like in Korean and unlike in English. This is predicted by the non-grammaticalized number hypothesis but not by the semantic competition hypothesis.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 31 / 36

Discussion

Discussion

Participants overwhelmingly said kohoXsg trials were true. This suggests they are assigning an exclusive interpretation to the plural, like in Korean and unlike in English. This is predicted by the non-grammaticalized number hypothesis but not by the semantic competition hypothesis. In other words, this is evidence that non-grammaticalization does not license semantic unmarkedness.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 31 / 36

Discussion

Caveat

Using the same materials with adult English speakers (n = 9), there is a preference for the exclusive interpretation.

kohoXSG 0.00 0.25 0.50 0.75 1.00

Frequency of response

false true kohoXSG 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II

Frequency of response

false true Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 32 / 36

Discussion

Caveat

Using the same materials with adult English speakers (n = 9), there is a preference for the exclusive interpretation.

kohoXSG 0.00 0.25 0.50 0.75 1.00

Frequency of response

false true kohoXSG 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II

Frequency of response

false true

However, the inclusive interpretation is still available ≈30% of the time, which is not the case for our participants.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 32 / 36

Discussion

Caveat

Using the same materials with adult English speakers (n = 9), there is a preference for the exclusive interpretation.

kohoXSG 0.00 0.25 0.50 0.75 1.00

Frequency of response

false true kohoXSG 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II

Frequency of response

false true

However, the inclusive interpretation is still available ≈30% of the time, which is not the case for our participants. This suggests that this isn’t transfer from the L1.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 32 / 36

Discussion

Caveat

Using the same materials with adult English speakers (n = 9), there is a preference for the exclusive interpretation.

kohoXSG 0.00 0.25 0.50 0.75 1.00

Frequency of response

false true kohoXSG 0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00 Study I Study II

Frequency of response

false true

However, the inclusive interpretation is still available ≈30% of the time, which is not the case for our participants. This suggests that this isn’t transfer from the L1. If anything, this just suggests it is worth investigating other things that influence the availability of the inclusive reading in English.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 32 / 36

Discussion

Conclusions

In a language with non-grammaticalized number, we should always expect to see exclusive interpretations of plural morphology.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 33 / 36

Discussion

Conclusions

In a language with non-grammaticalized number, we should always expect to see exclusive interpretations of plural morphology. An interesting question arises concerning the opposite: grammaticalized number.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 33 / 36

Discussion

Conclusions

In a language with non-grammaticalized number, we should always expect to see exclusive interpretations of plural morphology. An interesting question arises concerning the opposite: grammaticalized number.

◮ If grammaticalization necessitates that one value in a φ-feature paradigm

is semantically unmarked, then all languages with grammaticalized number should have an inclusive interpretation of plural morphology in (some) downward entailing contexts.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 33 / 36

Discussion

Conclusions

In a language with non-grammaticalized number, we should always expect to see exclusive interpretations of plural morphology. An interesting question arises concerning the opposite: grammaticalized number.

◮ If grammaticalization necessitates that one value in a φ-feature paradigm

is semantically unmarked, then all languages with grammaticalized number should have an inclusive interpretation of plural morphology in (some) downward entailing contexts.

◮ We are aware of no natural language counterexamples to this. Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 33 / 36

Discussion

Conclusions

In a language with non-grammaticalized number, we should always expect to see exclusive interpretations of plural morphology. An interesting question arises concerning the opposite: grammaticalized number.

◮ If grammaticalization necessitates that one value in a φ-feature paradigm

is semantically unmarked, then all languages with grammaticalized number should have an inclusive interpretation of plural morphology in (some) downward entailing contexts.

◮ We are aware of no natural language counterexamples to this. ◮ Might be worth doing an artificial language learning study, though . . . Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 33 / 36

Acknowledgments

Acknowledgements

Thanks to:

◮ The audience at the 2013 Workshop on the Acquisition of Quantification

at UMass Amherst.

◮ The MSU Language Acquisition Lab. ◮ Kait Ayres, Mayara Bitello, Karthik Durvasula, Brandon Grenier, Adam

Greene, Kenneth Hanson, Ashley Bartell Hesson, Taehoon Kim, Bee Leinbach, Alan Munn, John Sheets, and Heather Wiltse.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 34 / 36

References

References I

Fedzechkina, Maryia, T. Florian Jaeger, and Elissa L. Newport (2012). “Language learners restructure their input to facilitate efficient communication.” Proceedings of the National Academy of Sciences 109.44,

• pp. 17897–17902. doi: 10.1073/pnas.1215776109.

Heim, Irene (1991). “Artikel und Definitheit.” In: Semantik: Ein internationales Handbuch der zeitgenösischen Forschung. Ed. by Arnim von Stechow and Dieter Wunderlich. Berlin, Germany: Mouton de

• Gruyter. Chap. 22, pp. 487–535.

Kwon, SongNim and Anne Zribi-Hertz (2004). “Number from a syntactic perspective: Why plural marking looks ‘truer’ in French than in Korean.” In: Empirical Issues in Formal Syntax and Semantics 5. Ed. by O. Bonami and P. Cabredo Hofherr, pp. 133–158. Liter, Adam, Chrisopher C. Heffner, and Cristina Schmitt (2014). “The interpretation of plural morphology and grammaticalized number: an argument from artificial language learning.” Ms., Michigan State University, East Lansing, MI and University of Maryland, College Park, MD.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 35 / 36

References

References II

Sauerland, Uli (2008a). “Implicated Presuppositions.” In: The Discourse Potential of Underspecified Structures. Ed. by Anita Steube. Vol. 8. Language, Context and Cognition. Berlin, Germany: Walter de Gruyter,

• pp. 581–600. doi: 10.1515/9783110209303.4.581.

Sauerland, Uli (2008b). “On the Semantic Markedness of Phi-Features.” In: Phi-Theory: Phi-Features Across Modules and Interfaces. Ed. by Daniel Harbour, David Adger, and Susana Béjar. Vol. 16. Oxford Studies in Theoretical Linguistics. Oxford, UK: Oxford University Press. Chap. 3,

• pp. 57–82.

Sauerland, Uli, Jan Anderssen, and Kazuko Yatsushiro (2005). “The Plural Is Semantically Unmarked.” In: Linguistic Evidence: Empirical, Theoretical and Computational Perspectives. Ed. by Stephan Kepser and Marga Reis. Berlin, Germany: Mouton de Gruyter, pp. 409–430.

Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 36 / 36

Excluded participants

Not significantly above chance (33%) in final session on number interpretation of three different NP types:

◮ Study I: 1 participant ◮ Study II: 2 participants

Not significantly above chance (50%) in final session on verb negation:

◮ Study I: 3 participants ◮ Study II: 4 participants

Not significantly above chance (50%) in final session on object negation:

◮ Study I: 0 participants ◮ Study II: 3 participants

Not significantly above chance (50%) on non-critical (i.e., non-kohoXsg trials):

◮ Study I: 1 participant ◮ Study II: 2 participants

Final participant count for purposes of kohoXsg analysis:

◮ Study I: 17 − 1 − 3 − 0 − 1 = 12 ◮ Study II: 16 − 2 − 4 − 3 − 2 = 5 Liter et al. Non-grammaticalized number & plural October 17 – LCQ 2015 1 / 1