Phonological trends in the lexicon Michael Becker University of - - PowerPoint PPT Presentation

1 / 30

Phonological trends in the lexicon

Michael Becker University of Massachusetts Amherst michael.becker@phonologist.org

EVELIN 2012 UNICAMP / MIT Campinas, Brazil

Overview

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

2 / 30

• The empirical domain (today and Terça)
• Relations between words and morphophonology
• What are phonological trends in the lexicon?
• What do native speakers know?
• What can phonological trends teach us about grammar?
• Theory (Quarta)
• A grammatical model of lexical trends in Optimality Theory
• Grammatical vs. representational approaches (+rules)
• Problems with analogical models
• Practicum (Quinta)
• Finding lexical trends
• Designing, running and interpreting experiments
• Relating theoretical and experimental work

Morphophonology

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

3 / 30

Morphological relations

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

4 / 30

• English speakers understand that some words are related to

each other

• Some words are not related in the same way

Morphological relations

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

4 / 30

• English speakers understand that some words are related to

each other

• “dog” and “dogs” are related:

“Dogs” contains the sound and the meaning of “dog”.

• “Dogs” and “cats” are also related:
• they share the meaning “plural”
• they share a presence of a final alveolar strident —

[z] in [dOgz], [s] in [kæts].

• Some words are not related in the same way

Morphological relations

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

4 / 30

• English speakers understand that some words are related to

each other

• Some words are not related in the same way
• “Dog” and “cat” don’t share anything (beyond being nouns)
• “Scissors” does not have the singular “scissor”.

It is not a plural of anything.

• Similarly, “fairness” is not related to a plural

Morphological knowledge

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

5 / 30

• What do native speakers know about plurals?
• How do we know they know this?
• Why should we care?

Morphological knowledge

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

5 / 30

• What do native speakers know about plurals?
• They know singular words (“cat”, “dog”, “fairness”), and they

know plural words (“cats”, “dogs”, “scissors”).

• They know about connections between singulars and plurals:

[kæts]

PLURAL

← − − − → [kæt], [dOgz]

PLURAL

← − − − → [dOg], etc.

• They know about the connections between plural forms:

[s], [z], [Iz] all express plurality in [kæts], [dOgz], [sænwIÙIz] (also [aI] in [maIs], [laIs]).

• They know the distribution of the plural forms: the plural of

“car” is [kAôz], *[kAôs], *[kAôIz].

• How do we know they know this?
• Why should we care?

Morphological knowledge

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

5 / 30

• What do native speakers know about plurals?
• How do we know they know this?
• We test how they use real words (1):

“What word better describes this picture, ‘house’ or ‘houses’?”

• We test how they use real words (2):

“What is a better plural for ‘car’, [kAôz] or [kAôs]?”

• We test how they use made-up words (1):

“This is a wug. Now there is another one. There are two .”

• We test how they use made-up words (2):

“Is [splIks] a good description of this picture?” “Is [splIgz] a good description of this picture?”

• Why should we care?

Morphological knowledge

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

5 / 30

• What do native speakers know about plurals?
• How do we know they know this?
• Why should we care?
• Because children show that the task is far from trivial.

The wug test

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

6 / 30

• This technique was pioneered by Berko (1958)
• What did the kids do?
• Real words are different from novel words.

The wug test

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

6 / 30

• This technique was pioneered by Berko (1958)
• What did the kids do?
• Real words are different from novel words.

The wug test

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

6 / 30

• This technique was pioneered by Berko (1958)
• What did the kids do?
• The kids were very good at pluralizing wug (why?)
• but not very good at pluralizing niz
• At the same age, most of them were able to say and use the

plural glasses.

• Real words are different from novel words.

The wug test

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

6 / 30

• This technique was pioneered by Berko (1958)
• What did the kids do?
• Real words are different from novel words.

Possible explanations for the wug/niz difference:

• Something special about the distribution of [Iz] plurals?
• Something special about the phonological operation?
• Other?

Kids need time to master this aspect of the plural morphology. How do adults relate real words and nonce words?

Regular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

7 / 30

• Regular morpheme concatenation
• Regular concatenation with allomorphy
• Regular stem changes

Regular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

7 / 30

• Regular morpheme concatenation
• English progressive: talk ∼ talking, sleep ∼ sleeping, etc.

Completely exceptionless: have ∼ having, be ∼ being

• Arabic 1st person plural: [kitab-na] ‘our book’, [fihim-na] ‘we

understood’ or ‘he understood us’

• Turkish relativizer [-ki]
• Romance adverbs: [-mã] / [-mente] / [-m˜

eÙi]

• Regular concatenation with allomorphy
• Regular stem changes

Regular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

7 / 30

• Regular morpheme concatenation
• Regular concatenation with allomorphy
• English indefinite article: [@n] before vowels, [@] otherwise
• Berber reciprocal: [n-] if the base has a labial, e.g. [n-kaddab],

[m-] otherwise, e.g. [m-qarrad] Mostly regular:

• English plural: [Iz] after stridents, otherwise [s] after voiceless
• bstruents, [z] otherwise
• Portuguese plural: [s] after vowels, [is] otherwise
• Regular stem changes

Regular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

7 / 30

• Regular morpheme concatenation
• Regular concatenation with allomorphy
• Regular stem changes
• Chimwiini locative [-ni] (Kisseberth & Abasheikh 2004)

base noun locative Ùuwo Ùuwo:ni ‘book’ bu:ku buku:ni ‘book (western)’ qalbi qalbi:ni ‘heart’

• Slovenian adjectival [-@n] (Toporišiˇ

c 2000, Jurgec p.c.) base noun adjective ba"Rók ba"RóÙ-@n ‘Baroque’/‘baroque’

• "tR´

Ok

• "tRòÙ-@n

‘child’/‘childish’ "znák "znáÙ-@n ‘sign’/‘marked’ ["p´ @k@w] ‘hell’ / ["Ùúk-@ţ] ‘owl.DIM’

Irregular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

8 / 30

• Irregular morpheme concatenation
• Irregular allomorphy
• Irregular stem changes

Irregular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

8 / 30

• Irregular morpheme concatenation
• French feminine: [gKã-d] ‘big’, [blã-S] ‘white’, [gKi-z] ‘gray’,

[vEK-t] ‘green’, [du-s] ‘soft’, [p@ti-t] ‘small’

• Irregular allomorphy
• Irregular stem changes

Irregular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

8 / 30

• Irregular morpheme concatenation
• Irregular allomorphy
• Persian plural: [zan-An] ‘woman’ vs. [ketAb-hA] ‘book’,

[baÙe-hA] ‘child’

• Hebrew plural: ["giK ∼ gi"K-im] ‘chalk’ vs. ["kiK ∼ ki"K-ot] ‘wall’

(Becker 2009; Berent et al. 1999, 2002)

• Turkish aorist: [kapat-Wr] ‘close’, [sa"t-ar] ‘sell’, [da"l-ar] ‘knit’
• vs. [ka"l-Wr] ‘stay’ (Napiko˘

glu & Ketrez 2006; Becker 2009)

• Irregular stem changes

Irregular affixation

• Overview

Morphophonology

• Morphological

relations

• Morphological

knowledge

• The wug test
• Regular affixation
• Irregular affixation

Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

8 / 30

• Irregular morpheme concatenation
• Irregular allomorphy
• Irregular stem changes
• Portuguese plural: [saw ∼ sajs] ‘salt’ vs. [paw ∼ paws] ‘stick’

(Huback 2007; Gomes & Manoel 2010; Becker, Clemens & Nevins to appear)

• French plural: [Sakal ∼ Sakal] ‘jackal’ vs. [ZuKnal ∼ ZuKno]

‘newspaper’ (Becker, Clemens & Nevins to appeara)

• Spanish verbs: (Albright et al. 2001)

[mon"tar ∼ "monto] ‘mount’ [po"ðar ∼ "poðo] ‘prune’

• vs. [kon"tar ∼ "kwento] ‘count’

[po"ðer ∼ "pweðo] ‘be able’

• Russian genitive: ["kot ∼ ko"t-a] ‘cat’ vs. ["rot ∼ "rt-a] ‘mouth’

(Gouskova & Becker 2011)

• etc. etc. etc.

Lexical trends

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

9 / 30

Portuguese plurals

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

10 / 30

Becker, Clemens & Nevins (to appear).

http://becker.phonologist.org/projects/FrenchPortuguese/.

*Thanks to Filomena Sandalo for her help with this project. See also Huback (2007), Gomes & Manoel (2010).

• Predictable plurals after consonants, vowels, [j]
• Some final [w]’s stay faithful
• Some final [w]’s alternate with [j] (unfaithful to [back])
• A few final [w]’s alternate with [ë] (unfaithful to [lateral])
• Final [˜

@ ˜ w] is even crazier

Portuguese plurals

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

10 / 30

• Predictable plurals after consonants, vowels, [j]
• Nouns that end in [H] or [s] take the suffix form [-is]

floH floRis ‘flower’

flor

vO(j)s vOzis ‘voice’

voz

• Nouns that end in [j] or a vowel take [-s]

e"ROj e"ROjs ‘hero’

herói

"if˜ e "if˜ es ‘hyphen’

hífen

• No suffix on nouns that already end in an unstressed V + [s]

"lapis "lapis ‘pencil’

lápis

"õnibus "õnibus ‘bus’

ônibus

• Some final [w]’s stay faithful
• Some final [w]’s alternate with [j] (unfaithful to [back])
• A few final [w]’s alternate with [ë] (unfaithful to [lateral])
• Final [˜

@ ˜ w] is even crazier

Portuguese plurals

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

10 / 30

• Predictable plurals after consonants, vowels, [j]
• Some final [w]’s stay faithful

shape singular plural spelling mono "paw "paws

pau

‘stick’ "pnEw "pnEws

pneu

‘tire’ "gow "gows

gol

‘goal’ iamb ka."kaw ka."kaws

cacau

‘cocoa’ mu."zew mu."zews

museu

‘museum’ fu.Ùi."bow fu.Ùi."bows

futbol

‘football’ trochee "aw.kow "aw.kows

alcool

‘alcohol’

• Some final [w]’s alternate with [j] (unfaithful to [back])
• A few final [w]’s alternate with [ë] (unfaithful to [lateral])
• Final [˜

@ ˜ w] is even crazier

Portuguese plurals

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

10 / 30

• Predictable plurals after consonants, vowels, [j]
• Some final [w]’s stay faithful
• Some final [w]’s alternate with [j] (unfaithful to [back])

shape singular plural spelling mono "saw "sajs

sal

‘salt’ "mEw "mEjs

mel

‘honey’ "pROw "pROjs

prol

‘advantage’ iamb ZoH."naw ZoH."najs

jornal

‘newspaper’ a."nEw a."nEjs

anel

‘ring’ ba."hiw ba."his

barril

‘barrel’ trochee "ni.vew "ni.vejs

nível

‘level’ "hEp.Ùiw "hEp.Ùejs

réptil

‘reptile’

• A few final [w]’s alternate with [ë] (unfaithful to [lateral])
• Final [˜

@ ˜ w] is even crazier

Portuguese plurals

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

10 / 30

• Predictable plurals after consonants, vowels, [j]
• Some final [w]’s stay faithful
• Some final [w]’s alternate with [j] (unfaithful to [back])
• A few final [w]’s alternate with [ë] (unfaithful to [lateral])

mono "maw "maëis

mal

‘evil’ iamb ab"Riw ab"Riëis

abril

‘April’

• Final [˜

@ ˜ w] is even crazier

Portuguese plurals

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

10 / 30

• Predictable plurals after consonants, vowels, [j]
• Some final [w]’s stay faithful
• Some final [w]’s alternate with [j] (unfaithful to [back])
• A few final [w]’s alternate with [ë] (unfaithful to [lateral])
• Final [˜

@ ˜ w] is even crazier Let’s not worry about it today. See Huback (2007), Gomes & Manoel (2010).

The trend

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

11 / 30

• Distribution of plurals in the real words of the language:

0% 20% 40% 60% 80% 100%

mono iamb trochee

7 265 45 8 6 17 37 2

faithful intermediate alternating

• Nouns may or may not alternate, regardless of size
• There is a trend for more alternations in polysyllables
• Why should we care?

Trend productivity

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

12 / 30

• Variability / innovation
• Loanwords
• Wug-test

Trend productivity

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

12 / 30

• Variability / innovation
• Monosyllables move to faithful plurals:

["sajs] → ["saws] ‘salt’

• Polysyllables move to unfaithful plurals:

[de"gRaws] → [de"gRajs] ‘steps’ [Sa"pEws] → [Sa"pEjs] ‘hats’ [tRo"fEws] → [tRo"fEjs] ‘trophy’ “The rich get richer”

• Loanwords
• Wug-test

Trend productivity

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

12 / 30

• Variability / innovation
• Loanwords
• Monosyllables get faithful plurals:

["gow ∼ "gows] ‘goal’ ["ZEw ∼ "ZEws] ‘gel’

• Polysyllables get unfaithful plurals:

[koke"tEw ∼ koke"tEjs] ‘cocktail’

• What is the plural of skol?

[skow ∼ skows, skojs] [eskow ∼ eskows, eskojs]

• Wug-test

Trend productivity

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

12 / 30

• Variability / innovation
• Loanwords
• Wug-test
• 89 target items (47 monosyllables, 42 polysyllables)

["daw] [bi"ñaw] ["tOmew] ["gROw] [zu"gow] ["Sastow]

• Results:

faithful alternating mono trochee iamb

Interim summary

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

13 / 30

• Phonological trends are a part of the grammar
• These are not “exceptions”

Interim summary

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

13 / 30

• Phonological trends are a part of the grammar
• Phonological trends change the behavior of existing words
• Phonological trends affect loanwords
• Phonological trends extend productively to nonce words
• → they are a part of speaker’s knowledge
• → they are a part of a full analysis of the language
• These are not “exceptions”

Interim summary

• Overview

Morphophonology Lexical trends

• Portuguese plurals
• The trend
• Trend productivity
• Interim summary

Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

13 / 30

• Phonological trends are a part of the grammar
• These are not “exceptions”
• Which [w]-final words of Portuguese are exceptional?
• Both patterns get extended productively.
• The grammar allows multiple conflicting behaviors; the

grammar is irregular; the treatment of lexical items is not fully predictable.

Speakers’ knowledge

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

14 / 30

How much do speakers know?

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

15 / 30

• The evolutionary phonology/usage-based view (Blevins, Bybee,

Ohala, Wedel, others): speakers can learn any pattern given to them. Clearly wrong (we’ll see why soon).

• The universalist view (Becker, Nevins, others): speakers can only

learn universal/natural patterns. Probably correct (Related question: which patterns are universal/natural?)

• The natural bias view (Hayes, Zuraw, others): speakers are

biased to prefer universal/natural patterns (Related questions: which patterns are universal/natural? And how do speakers do both kinds of learning?)

Tagalog nasal substitution

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

16 / 30

Zuraw (2000)

• The alternation
• The lexical trend
• Extension to nonce words

Tagalog nasal substitution

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

16 / 30

Zuraw (2000)

• The alternation

Prefix causes initial obstruent to become nasal — in some words, but not others b mag-bi"gaj

‘to give’

ma-mi"gaj

‘to distribute’

big"kas

‘pronouncing’

mam-bi-big"kas

‘reciter’

d da"la:Nin

‘prayer’

Pa-pi-na"laNin

‘to pray’

di"nig

‘audible’

pan-di"nig

‘sense of hearning’

• The lexical trend
• Extension to nonce words

Tagalog nasal substitution

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

16 / 30

Zuraw (2000)

• The alternation
• The lexical trend

pa!- (noun)

27 20 7

1

25 11

8 7 18 13 5 27 17 26 3 8 6 7 0% 20% 40% 60% 80% 100% p t s k b d g Unsubstituted Varies Substituted

• Extension to nonce words

Tagalog nasal substitution

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

16 / 30

Zuraw (2000)

• The alternation
• The lexical trend
• Extension to nonce words
• 5
• 4
• 3
• 2
• 1

1 2 3 4

p t s k b d g

Dutch voicing

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

17 / 30

Ernestus & Baayen (2003)

• The alternation
• Lexicon (CELEX, Baayen et al. 1995) & nonce words

Dutch voicing

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

17 / 30

Ernestus & Baayen (2003)

• The alternation

Voiceless obstruents word-finally (“final devoicing”). Before a vowel, obstruents become voiced in some words, but not others. imperative infinitive [vErVEit] [vErVEit@n] ‘reproach’ faithful [vErVEit] [vErVEid@n] ‘widen’ alternating

• Lexicon (CELEX, Baayen et al. 1995) & nonce words

Dutch voicing

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

17 / 30

Ernestus & Baayen (2003)

• The alternation
• Lexicon (CELEX, Baayen et al. 1995) & nonce words

P T S F X P T S F X lexical data experimental data % [+voice]

40 80

% [+voice]

40 80

The effect of place in the lexicon is very different from Tagalog, but speakers learn it equally well.

Turkish voicing

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

18 / 30

Becker, Ketrez & Nevins (2011)

• The phenomenon
• The lexical trend (place and size)
• Extension to nonce words

Turkish voicing

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

18 / 30

Becker, Ketrez & Nevins (2011)

• The phenomenon

Voiceless obstruents word-finally (“final devoicing”). Before a vowel, obstruents become voiced in some words, but not others. nominative accusative [anaÙh] [anaÙh-1] ‘female cub’ faithful [amaÙh] [amaÃ-1] ‘target’ alternating

• The lexical trend (place and size)
• Extension to nonce words

Turkish voicing

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

18 / 30

Becker, Ketrez & Nevins (2011)

• The phenomenon
• The lexical trend (place and size)

DOR PAL COR LAB CVC CVCC CVCVC

alternating vacillating non-alternating

• Extension to nonce words

Turkish voicing

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

18 / 30

Becker, Ketrez & Nevins (2011)

• The phenomenon
• The lexical trend (place and size)
• Extension to nonce words

DOR PAL COR LAB CVC CVCC CVCVC

Interim summary

• Overview

Morphophonology Lexical trends Speakers’ knowledge

• How much do

speakers know?

• Tagalog nasal

substitution

• Dutch voicing
• Turkish voicing
• Interim summary

Limits of speakers’ knowledge Is grammar not enough? Summary

19 / 30

• Speakers track trends (lexical statistics) for each place of

articulation

• Each place of articulation is tracked separately
• Turkish: Speakers track trends for each place within each

phonological size

• Humans are good statistical learners

“LAW OF FREQUENCY MATCHING Speakers of languages with variable lexical patterns respond stochastically when tested on such patterns. Their responses aggregately match the lexical frequencies.” (Hayes et al. 2009)

Limits of speakers’ knowledge

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

20 / 30

Turkish again

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

21 / 30

Becker, Ketrez & Nevins (2011)

• Lexicon: more alternations after back vowels
• Experiment: no effect
• Our explanation: universal bias

Turkish again

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

21 / 30

Becker, Ketrez & Nevins (2011)

• Lexicon: more alternations after back vowels

+back !back

• Experiment: no effect
• Our explanation: universal bias

Turkish again

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

21 / 30

Becker, Ketrez & Nevins (2011)

• Lexicon: more alternations after back vowels
• Experiment: no effect

+back !back

• Our explanation: universal bias

Turkish again

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

21 / 30

Becker, Ketrez & Nevins (2011)

• Lexicon: more alternations after back vowels
• Experiment: no effect
• Our explanation: universal bias
• Phonetics: Vowel backness is not related to voicing
• Typology: no known interactions of backness and voicing
• Speakers only track statistics that are motivated by universal

grammar Surfeit of the stimulus = speakers ignore data if their grammar isn’t set up to capture it. This is a problem for linguists who think that humans can learn any kind of pattern.

English plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

22 / 30

Becker, Nevins & Levine (to appear)

• The alternation
• The lexical trend (dictionary study)
• The lexical trend (experimentally)
• Extension to nonce words

English plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

22 / 30

Becker, Nevins & Levine (to appear)

• The alternation

singular plural naIf naIvz, *naIfs ‘knife’ alternating pæT pæðz, *pæTs ‘path’ stIf stIfs, *stIvz ‘stiff’ faithful dET dETs, *dEðz ‘death’

• The lexical trend (dictionary study)
• The lexical trend (experimentally)
• Extension to nonce words

English plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

22 / 30

Becker, Nevins & Levine (to appear)

• The alternation
• The lexical trend (dictionary study)

CMU dictionary: 126 nouns [f/T]-final nouns Monosyllables: 28% alternate Polysyllables: no alternatiosd, 100% faithful The opposite of Portuguese, Turkish, French, Russian...

• The lexical trend (experimentally)
• Extension to nonce words

English plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

22 / 30

Becker, Nevins & Levine (to appear)

• The alternation
• The lexical trend (dictionary study)
• The lexical trend (experimentally)

200 speakers rated our 126 items (Amazon’s Mechanical Turk)

voiceless voiced mono iamb trochee

• Extension to nonce words

English plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

22 / 30

Becker, Nevins & Levine (to appear)

• The alternation
• The lexical trend (dictionary study)
• The lexical trend (experimentally)

200 speakers rated our 126 items (Amazon’s Mechanical Turk)

voiceless voiced mono iamb trochee

stiff ref brief serf cuff strength sniff bluff graph breath prof month deathcough growth chef cliff goof

• af

puff clef staff gaffe nymph myth laugh skiff quaff fife proof safe heath turf waif whiff spoof sloth tiff reef gulf faith fourth hearth earth chief trough berth roof broth booth Goth width sleuth youthcloth truth swath

• ath

bath wreath mouth wraithmoth wharf sheath path sheaf knife dwarf loaf wife thief wolf leaf selfhoof half scarf shelf calf elf life belief earmuff midriff eighteenth handcuffparagraph castoff blacksmith motif carafe pilaf epitaph sabertooth polymath aftermath coelacanth giraffe relief psychopath behalf vermouth hieroglyph goliath sheriff triumph tariff caliph pontiff azimuth serif bailiff monolith mastiff plaintiff hyacinth absinth billionth eightieth mammoth seraph zenith Sabbath kerchief Behemoth

• Extension to nonce words

English plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

22 / 30

Becker, Nevins & Levine (to appear)

• The alternation
• The lexical trend (dictionary study)
• The lexical trend (experimentally)
• Extension to nonce words

voiceless voiced mono iamb trochee

Interim summary

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

23 / 30

• Observation: monosyllables can’t be less faithful than

polysyllables

• Our explanation: surfeit of the stimulus

Interim summary

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

23 / 30

• Observation: monosyllables can’t be less faithful than

polysyllables

• Portuguese, Turkish, French, Russian lexicon: monosyllables

more faithful

• Portuguese, Turkish, French, Russian nonce words:

monosyllables more faithful

• English lexicon: monosyllables less faithful
• English nonce words: monosyllables just as faithful as

polysyllables

• Our explanation: surfeit of the stimulus

Interim summary

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

23 / 30

• Observation: monosyllables can’t be less faithful than

polysyllables

• Our explanation: surfeit of the stimulus

Speakers can’t learn that monosyllables alternate more than polysyllables Preview of our Optimality Theory analysis:

• Faith ≫ Markedness ≫ Faith-mono

Faith ≫ Faith-mono ≫ Markedness

• Markedness ≫ Faith ≫ Faith-mono

Markedness ≫ Faith-mono ≫ Faith

• Faith-mono ≫ Markedness ≫ Faith

If we assume Universal Grammar doesn’t have a constraint that protects polysyllables, then we get the correct range of languages.

Beyond wug-testing

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

24 / 30

• Artificial Language experiment (Becker, Nevins & Levine)
• Result: Monosyllables protected
• Result: Monosyllables protected (stats)
• Conclusion

Beyond wug-testing

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

24 / 30

• Artificial Language experiment (Becker, Nevins & Levine)

monosyllabic training iambic training Training 10 stop-final monos 10 stop-final iambs "mip "mibni t@"gep t@"gebni "stut "studni g@"Sut g@"Sudni 5 sonorant-finals: "muN-ni, n@"Ãol-ni Testing 10 stop-final monos 10 stop-final monos "gaIp "gaIp "klet "klet 10 stop-final iambs 10 stop-final iambs f@"Ùop f@"Ùop b@"git b@"git 10 sonorant-finals: "pler, Z@"taIm

• Result: Monosyllables protected
• Result: Monosyllables protected (stats)
• Conclusion

Beyond wug-testing

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

24 / 30

• Artificial Language experiment (Becker, Nevins & Levine)
• Result: Monosyllables protected

Monosyllabic training group

voiceless voiced

mono iamb Iambic training group

voiceless voiced

mono iamb

• Result: Monosyllables protected (stats)
• Conclusion

Beyond wug-testing

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

24 / 30

• Artificial Language experiment (Becker, Nevins & Levine)
• Result: Monosyllables protected
• Result: Monosyllables protected (stats)

What is the probability of an alternating response given that the response ... ... came from a participant in the monosyllabic training group? ... came from a participant in the iambic training group? ... was to an item of the shape seen in training? ... was to an item of the shape not seen in training?

β

SE(β) z p-value (Intercept) .49 .32 1.59

>.1

group

−.28

.31

−.92 >.1

untrained

−.30

.07

−4.16 <.0001

group:untrained

−.21

.08

−2.79 <.01

• Conclusion

Beyond wug-testing

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge

• Turkish again
• English plurals
• Interim summary
• Beyond wug-testing

Is grammar not enough? Summary

24 / 30

• Artificial Language experiment (Becker, Nevins & Levine)
• Result: Monosyllables protected
• Result: Monosyllables protected (stats)
• Conclusion
• When given a chance, English speakers prefer a

Portuguese-style language.

• What do English speakers know?
• Lexicon: many alternating monosyllables, few/no

alternating polysyllables.

• Nonce words: alternation applied to monosyllables and

polysyllables equally.

• New alternation/language: monosyllables are protected.
• Speakers do not treat all patterns equally.
• Explanation: Grammar filters the lexical trends.

Is grammar not enough?

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

25 / 30

Hebrew plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

26 / 30

• Allomorph selection: masculines often take [–im], rarely [–ot]
• Lexicon: more [–ot] after masculines with [o]
• Experiment (Becker 2009)
• Experiment (Berent et al. 1999, 2002)
• Analogy? Lexical magnet(?)

Hebrew plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

26 / 30

• Allomorph selection: masculines often take [–im], rarely [–ot]

singular plural majority pattern "giK gi"Kim ‘chalk’ a"lon alo"nim ‘oak’ minority pattern "kiK ki"Kot ‘wall’ Xa"lon Xalo"not ‘window’

• Lexicon: more [–ot] after masculines with [o]
• Experiment (Becker 2009)
• Experiment (Berent et al. 1999, 2002)
• Analogy? Lexical magnet(?)

Hebrew plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

26 / 30

• Allomorph selection: masculines often take [–im], rarely [–ot]
• Lexicon: more [–ot] after masculines with [o]

f a e io u m a e i o u

im

• t

8822 nouns from a dictionary (Bolozky & Becker 2006)

• Experiment (Becker 2009)
• Experiment (Berent et al. 1999, 2002)
• Analogy? Lexical magnet(?)

Hebrew plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

26 / 30

• Allomorph selection: masculines often take [–im], rarely [–ot]
• Lexicon: more [–ot] after masculines with [o]
• Experiment (Becker 2009)
• More [ot] when stem contains [o]
• Experiment (Berent et al. 1999, 2002)
• Analogy? Lexical magnet(?)

Hebrew plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

26 / 30

• Allomorph selection: masculines often take [–im], rarely [–ot]
• Lexicon: more [–ot] after masculines with [o]
• Experiment (Becker 2009)
• Experiment (Berent et al. 1999, 2002)
• Items designed based on real [ot]-takers:

real word same place different place many features

SikoK (-im) SigoK SiboK midov ţinoK (-ot) ţiloK ţikoK bikov

• Most [ot] with nouns that are almost identical to real words
• Significantly less [ot] with moderate similarity
• Berent’s interpretation: [ot]-takers subject to analogy, [im]

affixed by rule (Pinker’s dual model)

• Analogy? Lexical magnet(?)

Hebrew plurals

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

26 / 30

• Allomorph selection: masculines often take [–im], rarely [–ot]
• Lexicon: more [–ot] after masculines with [o]
• Experiment (Becker 2009)
• Experiment (Berent et al. 1999, 2002)
• Analogy? Lexical magnet(?)

Berent et al. (1999, 2002) assume that nonce words take [ot] due to analogy, including nouns that only resemble [ot] takers just by having an [o]. But that’s different from overall similarity. ţinoK ←

→ midov (takes [ot] by analogy...?)

ţinoK ←

→ ţinaK (analogy doesn’t work...?)

My interpretation:

• selection of [ot] by a grammar (vowel harmony)
• “lexical magnets” (cf. perceptual magnet) — a real word pulls

stronger than the grammar = analogy. More against analogy in Albright & Hayes (2003).

Hungarian genitive

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

27 / 30

Another case of allomorph selection (Hayes & Londe 2006; Hayes et al. 2009)

• Lexical exceptions after front unrounded vowels [i:, e:, i, E]
• Natural factors: vowel height, number of front unrounded vowels
• Unnatural factors: final consonants, clusters
• Extension to nonce words
• Hayes et al.’s interpretation: naturalness bias
• My response(?)

Hungarian genitive

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

27 / 30

Another case of allomorph selection (Hayes & Londe 2006; Hayes et al. 2009)

• Lexical exceptions after front unrounded vowels [i:, e:, i, E]

ţi:m ţi:m-nEk ‘address’ hi:d hi:d-nOk ‘bridge’

• Natural factors: vowel height, number of front unrounded vowels
• Unnatural factors: final consonants, clusters
• Extension to nonce words
• Hayes et al.’s interpretation: naturalness bias
• My response(?)

Hungarian genitive

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

27 / 30

Another case of allomorph selection (Hayes & Londe 2006; Hayes et al. 2009)

• Lexical exceptions after front unrounded vowels [i:, e:, i, E]
• Natural factors: vowel height, number of front unrounded vowels
• Most [nEk] after [E], less after [e:], least after [i, i:]
• More [nEk] after two front unrounded vowels: [kErt] ‘garden’
• vs. [rEpEs] ‘splint’
• Unnatural factors: final consonants, clusters
• Extension to nonce words
• Hayes et al.’s interpretation: naturalness bias
• My response(?)

Hungarian genitive

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

27 / 30

Another case of allomorph selection (Hayes & Londe 2006; Hayes et al. 2009)

• Lexical exceptions after front unrounded vowels [i:, e:, i, E]
• Natural factors: vowel height, number of front unrounded vowels
• Unnatural factors: final consonants, clusters
• More [nEk] after labials [p, b, m] (the opposite of labial

attraction!)

• More [nEk] after sibilants [s, z, S, Z, ţ, Ù, Ã] (makes no sense)
• More [nEk] after clusters (makes no sense)
• Extension to nonce words
• Hayes et al.’s interpretation: naturalness bias
• My response(?)

Hungarian genitive

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

27 / 30

Another case of allomorph selection (Hayes & Londe 2006; Hayes et al. 2009)

• Lexical exceptions after front unrounded vowels [i:, e:, i, E]
• Natural factors: vowel height, number of front unrounded vowels
• Unnatural factors: final consonants, clusters
• Extension to nonce words
• Speakers extend the natural factors
• Speakers also extend the unnatural factors, but more weakly
• Hayes et al.’s interpretation: naturalness bias
• My response(?)

Hungarian genitive

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

27 / 30

Another case of allomorph selection (Hayes & Londe 2006; Hayes et al. 2009)

• Lexical exceptions after front unrounded vowels [i:, e:, i, E]
• Natural factors: vowel height, number of front unrounded vowels
• Unnatural factors: final consonants, clusters
• Extension to nonce words
• Hayes et al.’s interpretation: naturalness bias
• Hayes et al.: speakers can learn any pattern, but they prefer

natural patterns

• Hayes et al.: Becker & Nevins are too radical
• My response(?)

Hungarian genitive

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough?

• Hebrew plurals
• Hungarian genitive

Summary

27 / 30

Another case of allomorph selection (Hayes & Londe 2006; Hayes et al. 2009)

• Lexical exceptions after front unrounded vowels [i:, e:, i, E]
• Natural factors: vowel height, number of front unrounded vowels
• Unnatural factors: final consonants, clusters
• Extension to nonce words
• Hayes et al.’s interpretation: naturalness bias
• My response(?)
• Maybe many items in Hayes et al.’s experiment were “lexical

magnets” (needs to be tested)

• Maybe the trends speakers track for allomorph selection

(Hungarian, Hebrew) = trends speakers track for stem changes (Portuguese, English, etc.)

• Question: what kind of theory allows both natural and

unnatural factors, and prefers natural factors?

Summary

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

28 / 30

Nonce words are an effect of grammar

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

29 / 30

• In irregular morphophonology, speakers track lexical statistics
• Detailed statistical tracking: place, size
• Limits: protection of monosyllables
• Limits: unnatural factors
• Theory?

Nonce words are an effect of grammar

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

29 / 30

• In irregular morphophonology, speakers track lexical statistics
• “Law of Frequency Matching”
• No notion of exception
• Detailed statistical tracking: place, size
• Limits: protection of monosyllables
• Limits: unnatural factors
• Theory?

Nonce words are an effect of grammar

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

29 / 30

• In irregular morphophonology, speakers track lexical statistics
• Detailed statistical tracking: place, size
• Speakers track each place separately (Tagalog, Dutch,

Turkish)

• Speakers track each place within each size (Turkish)
• Limits: protection of monosyllables
• Limits: unnatural factors
• Theory?

Nonce words are an effect of grammar

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

29 / 30

• In irregular morphophonology, speakers track lexical statistics
• Detailed statistical tracking: place, size
• Limits: protection of monosyllables
• Portuguese, French, Russian, etc.: monosyllables protected
• English lexicon: monos less protected than polys
• English nonce words: equal protection for monos and polys
• English speakers, artificial grammar: monos protected
• English speakers can’t learn the counter-typological trend
• Limits: unnatural factors
• Theory?

Nonce words are an effect of grammar

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

29 / 30

• In irregular morphophonology, speakers track lexical statistics
• Detailed statistical tracking: place, size
• Limits: protection of monosyllables
• Limits: unnatural factors
• Turkish lexicon: more voicing after back vowels
• Turkish nonce words: equal voicing after back and front

vowels

• Hungarian lexicon: more [nEk] after sibilants
• Hungarian nonce words: a little bit more [nEk] after sibilants
• Requirement/preference for natural patterns
• Hebrew: lexical magnets(?)
• Theory?

Nonce words are an effect of grammar

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

29 / 30

• In irregular morphophonology, speakers track lexical statistics
• Detailed statistical tracking: place, size
• Limits: protection of monosyllables
• Limits: unnatural factors
• Theory?
• Speakers track lexical statistics → traditional generative

grammar is too strict

• Speakers track natural lexical statistics → pure statistics are

too permissive

• Solution: statistics + generative grammar

References

• Overview

Morphophonology Lexical trends Speakers’ knowledge Limits of speakers’ knowledge Is grammar not enough? Summary

• Nonce words are an

effect of grammar

• References

30 / 30

Albright, Adam, Argelia Andrade & Bruce Hayes (2001). Segmental environments of Spanish diphthongization. In Adam Albright & Taehong Cho (eds.) UCLA Working Papers in Linguistics 7 (Papers in Phonology 5), UCLA. 117–151. Albright, Adam & Bruce Hayes (2003). Rules vs. Analogy in English past tenses: a computational/experimental study. Cognition 90. 119–161. Baayen, R. Harald, Richard Piepenbrock & Leon Gulikers (1995). The CELEX Lexical Database (CD-ROM). Linguistic Data Consortium, University of Pennsylvania, Philadelphia, PA. Becker, Michael (2009). Phonological Trends in the Lexicon: The Role of

• Constraints. Ph.D. dissertation, University of Massachusetts Amherst.

Becker, Michael, Lauren Eby Clemens & Andrew Nevins (to appeara). The monosyllabicity effect in French and Brazilian Portuguese. Natural Language and Linguistic Theory Ms. University of Massachusetts Amherst. Becker, Michael, Nihan Ketrez & Andrew Nevins (2011). The surfeit of the stimulus: Analytic biases filter lexical statistics in Turkish laryngeal

• alternations. Language 87. 84–125.