Phonological trends in the lexicon Theory Michael Becker - - PowerPoint PPT Presentation

1 / 18

Phonological trends in the lexicon — Theory

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

EVELIN 2012 MIT / UNICAMP Campinas, Brazil

Overview

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

2 / 18

• Basic mechanics of Optimality Theory (OT, Prince & Smolensky

1993/2004)

• Ranking arguments
• A grammatical approach to lexical trends
• Learning lexical trends
• Projecting trends onto novel words
• Representational vs. grammatical approaches
• Surface-true UR’s vs. abstract UR’s
• Representational vs. grammatical = OT vs. rules

OT Basics

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

3 / 18

Predictable distibutions

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

4 / 18

• Complementary distribution
• OT (Prince & Smolensky 1993/2004): markedness constraints
• Factorial typology

Predictable distibutions

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

4 / 18

• Complementary distribution
• [sa, Si, su], *[Sa, si, Su]

(e.g. native vocabulary of Japanese)

• Two generalizations:
• nly [S] appears before [i], only [s] appears elsewhere
• Rule-based analysis:
• One phoneme: /s/, no S in underlying representations
• One rule: [s] → [−ant] /

i

• rule application:

Underlying representation (UR) / susi / Rule application S Surface representation (SR, PR) [ suSi ]

• No /suSi/ in this language
• OT (Prince & Smolensky 1993/2004): markedness constraints
• Factorial typology

Predictable distibutions

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

4 / 18

• Complementary distribution
• OT (Prince & Smolensky 1993/2004): markedness constraints

/si/ *si *S a. si *!

• b. ☞ Si

* /sa/ *si *S

• a. ☞ sa

b. Sa *! /Si/ *si *S a. si *!

• b. ☞ Si

* /Sa/ *si *S

• a. ☞ sa

b. Sa *! No phonemes = all sounds are allowed in URs.

• Factorial typology

Predictable distibutions

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

4 / 18

• Complementary distribution
• OT (Prince & Smolensky 1993/2004): markedness constraints
• Factorial typology

/si/ *S *si

• a. ☞ si

* b. Si *! /sa/ *S *si

• a. ☞ sa

b. Sa *! /Si/ *S *si

• a. ☞ si

* b. Si *! /Sa/ *S *si

• a. ☞ sa

b. Sa *! What kind of language did we get?

Contrast

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

5 / 18

• English contrasts [su] ‘sue’ and [Su] ‘shoe’
• English also contrasts [si] ‘sea’ and [Si] ‘she’
• English does not contrast [Sr] ‘shrew’, ‘shriek’, etc. and *[sr]

Contrast

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

5 / 18

• English contrasts [su] ‘sue’ and [Su] ‘shoe’
• We need to be faithful to [±ant].
• IDENT(ant): Assign one violation mark to every output

segment that has an input correspondent, and where the [±ant] feature of the input doesn’t match the [±ant] feature of the output /su/ *si IDENT(ant) *S

• a. ☞ su

b. Su *! * /Su/ *si IDENT(ant) *S a. su *!

• b. ☞ Su

*

• English also contrasts [si] ‘sea’ and [Si] ‘she’

Contrast

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

5 / 18

• English contrasts [su] ‘sue’ and [Su] ‘shoe’
• English also contrasts [si] ‘sea’ and [Si] ‘she’
• IDENT(ant) ≫ *si

/si/ IDENT(ant) *si *S

• a. ☞ si

* b. Si *! * /Si/ IDENT(ant) *si *S a. si *! *

• b. ☞ Si

*

• English does not contrast [Sr] ‘shrew’, ‘shriek’, etc. and *[sr]

Contrast

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

5 / 18

• English contrasts [su] ‘sue’ and [Su] ‘shoe’
• English also contrasts [si] ‘sea’ and [Si] ‘she’
• English does not contrast [Sr] ‘shrew’, ‘shriek’, etc. and *[sr]

/sri/, /Sri/ *sr IDENT(ant) *si *S

• a. ☞ Sri

(*) * b. sri *! (*) The [s]—[S] contrast is neutralized before [r] Compare with Arabic [Sr1b] ‘drink’ vs. [sr1q] ‘steal’

Sources of constraints

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

6 / 18

• Prince & Smolensky (1993/2004): all constraints are innate
• Hayes & Wilson (2008): induction of markedness constraints = all

markedness constraints are language-specific

• As of now, nobody tried to induce faithfulness constraints.

Sources of constraints

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

6 / 18

• Prince & Smolensky (1993/2004): all constraints are innate
• Not 100% true — Prince & Smolensky (1993/2004) also had

constraints that align specific morphemes with word edges, e.g. Tagalog’s ALIGN(um-)

• Prince & Smolensky (1993/2004) weren’t worried about a

huge Universal Grammar

• Hayes & Wilson (2008): induction of markedness constraints = all

markedness constraints are language-specific

• As of now, nobody tried to induce faithfulness constraints.

Sources of constraints

• Overview

OT Basics

• Predictable

distibutions

• Contrast
• Sources of

constraints Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

6 / 18

• Prince & Smolensky (1993/2004): all constraints are innate
• Hayes & Wilson (2008): induction of markedness constraints = all

markedness constraints are language-specific

• Markedness constraints are generated by the learner, and

kept if useful in making the data probable (what does that mean?)

• Universal Grammar is small, but important: features,

constraint templates, autosegmental tiers

• No discussion of faithfulness
• (How does this prevent learning of unnatural trends in

Turkish, or the weaker learning of unnatural trends in Hungarian?)

• As of now, nobody tried to induce faithfulness constraints.

Rankings

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

7 / 18

Ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

8 / 18

• Is a ranking needed?
• Comparative tableau 1 (Prince 2002)
• Comparative tableau 2
• Comparative tableau 3
• Summary:

Ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

8 / 18

• Is a ranking needed?

input C1 C2

• a. ☞ output1

b.

• utput2

* input C1 C2

• a. ☞ output1

* b.

• utput2

* * input C1 C2

• a. ☞ output1

* *** b.

• utput2

** * input C1 C2 C3

• a. ☞ output1

* b.

• utput2

* *

• Comparative tableau 1 (Prince 2002)
• Comparative tableau 2

Comparative tableau 3

Ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

8 / 18

• Is a ranking needed?
• Comparative tableau 1 (Prince 2002)

Standard tableau: input C1 C2 C3

• a. ☞ output1

* b.

• utput2

* * Comparative tableau: input C1 C2 C3

• utput1 ≻ output2

W W L Rule: every L must be dominated by some W

• Comparative tableau 2
• Comparative tableau 3
• Summary:

Ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

8 / 18

• Is a ranking needed?
• Comparative tableau 1 (Prince 2002)
• Comparative tableau 2

Standard tableau: input C1 C2 C3

• a. ☞ output1

* * b.

• utput2

* * Comparative tableau: input C1 C2 C3

• utput1 ≻ output2

W L One W is more informative than 2 W’s

• Comparative tableau 3
• Summary:

Ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

8 / 18

• Is a ranking needed?
• Comparative tableau 1 (Prince 2002)
• Comparative tableau 2
• Comparative tableau 3

Standard tableau: input C1 C2 C3

• a. ☞ output1

* * b.

• utput2

* Comparative tableau: input C1 C2 C3

• utput1 ≻ output2

W L L Two L ’s are more informative than one L

• Summary:

Ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

8 / 18

• Is a ranking needed?
• Comparative tableau 1 (Prince 2002)
• Comparative tableau 2
• Comparative tableau 3
• Summary:
• Comparative tableaux identify ranking arguments
• Every L must be dominated by some W

More ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

9 / 18

• Combining ranking arguments
• Fusion
• Rules of fusion
• Recursive constraint “demotion” (RCD, Tesar 1995, 1996;

Tesar & Smolensky 1998; Prince & Tesar 2004)

More ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

9 / 18

• Combining ranking arguments

Standard tableaux: input1 C1 C2 C3

• a. ☞ output1

* b.

• utput2

* * input2 C1 C2 C3

• a. ☞ output3

* b.

• utput4

* Comparative tableau: C1 C2 C3 /input1/ output1 ≻ output2 W W L /input2/ output3 ≻ output4 L W

• Fusion

More ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

9 / 18

• Combining ranking arguments
• Fusion

C1 C2 C3 /input1/ output1 ≻ output2 W W L /input2/ output3 ≻ output4 L W From row 1: Either C1 ≫ C3 or C2 ≫ C3 (or both) From row 2: C3 ≫ C2 Fusion: C1 C2 C3 W L L From fusion: C1 ≫ C2, and C1 ≫ C3 From row 2: C3 ≫ C2 Final ranking: C1 ≫ C3 ≫ C2

• Rules of fusion
• Recursive constraint “demotion” (RCD, Tesar 1995, 1996;

More ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

9 / 18

• Combining ranking arguments
• Fusion
• Rules of fusion
• W + L = L
• W + e = W
• L + e = L
• W + W = W
• L + L = L
• e + e = e
• Recursive constraint “demotion” (RCD, Tesar 1995, 1996;

Tesar & Smolensky 1998; Prince & Tesar 2004)

More ranking arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

9 / 18

• Combining ranking arguments
• Fusion
• Rules of fusion
• Recursive constraint “demotion” (RCD, Tesar 1995, 1996;

Tesar & Smolensky 1998; Prince & Tesar 2004)

• Identify column(s) with no L

’s, install

• Remove rows that get W’s from the installed constraint(s)
• Continue until all constraints are installed
• C1

C2 C3 /input1/ output1 ≻ output2 W W L /input2/ output3 ≻ output4 L W

• From language data to a ranking, in a principled way
• Easy to implement computationally (OT-Soft, OT-Help, Praat)

Conflicting arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

10 / 18

• Conflict
• Identifying conflict
• Conflict resolution with cloning (Pater 2006, 2008)

Conflicting arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

10 / 18

• Conflict

Standard tableaux: input1 C1 C2

• a. ☞ output1

* b.

• utput2

* input2 C1 C2

• a. ☞ output3

* b.

• utput4

* Comparative tableau: C1 C2 /input1/ output1 ≻ output2 W L /input2/ output3 ≻ output4 L W

• Identifying conflict

Conflicting arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

10 / 18

• Conflict
• Identifying conflict

Comparative tableau: C1 C2 /input1/ output1 ≻ output2 W L /input2/ output3 ≻ output4 L W No columns without L ’s → constraints left uninstalled. RCD is stuck. Tesar’s conclusion: something is wrong in the analysis.

• Conflict resolution with cloning (Pater 2006, 2008)

Conflicting arguments

• Overview

OT Basics Rankings

• Ranking arguments
• More ranking

arguments

• Conflicting arguments

Trends with constraint cloning Grammatical vs. respresentational approaches

10 / 18

• Conflict
• Identifying conflict
• Conflict resolution with cloning (Pater 2006, 2008)

Pater: the language needs both rankings C1 C2 /input1/ output1 ≻ output2 W L /input2/ output3 ≻ output4 L W Cloning and lexical indexation: C1input1 C1input2 C2 /input1/ output1 ≻ output2 W L /input2/ output3 ≻ output4 L W Now RCD can find a ranking!

Trends with constraint cloning

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

11 / 18

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• Standard tableaux:
• Comparative tableau:
• Install *[σw
• Clone IDENT
• Grammar with lexical information
• And as more items are added...

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• After vowels, [is] creates an ONSET violation

baHkus ≻ baHkuis

• After consonants, [s] creates a *COMPLEX violation

floRis ≻ floRs

• After[j], [is] creates an OCP violation

eROjis ≻ eROjs

• With [w], [is] is out
• Standard tableaux:
• Comparative tableau:
• Install *[σw
• Clone IDENT
• Grammar with lexical information
• And as more items are added...

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• *[sawis], *[pawis], *[anEwis], *[muzewis]
• Presumably *[σw
• [s] is okay for some words: [paws], [muzews]
• Other words don’t allow [ws]: *[saws], *[anEws]
• The change from [w] to [j] creates a violation of IDENT(back)
• Standard tableaux:
• Comparative tableau:
• Install *[σw
• Clone IDENT
• Grammar with lexical information
• And as more items are added...

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• Standard tableaux:

/ paw + (i)s / *[σw *ws IDENT a. pawis *

• b. ☞ paws

* c. pajs * / saw + (i)s / *[σw *ws IDENT a. sawis * b. saws *

• c. ☞ sajs

*

• Comparative tableau:
• Install *[σw
• Clone IDENT

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• Standard tableaux:
• Comparative tableau:

*[σw *ws IDENT /paw/ paws ≻ pawis W L /paw/ paws ≻ pajs L W /saw/ sajs ≻ sawis W L /saw/ sajs ≻ saws W L

• Install *[σw
• Clone IDENT
• Grammar with lexical information
• And as more items are added...

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• Standard tableaux:
• Comparative tableau:
• Install *[σw

*ws IDENT /paw/ paws ≻ pajs L W /saw/ sajs ≻ saws W L

• Clone IDENT
• Grammar with lexical information
• And as more items are added...

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• Standard tableaux:
• Comparative tableau:
• Install *[σw
• Clone IDENT

*ws IDENTpaw IDENTsaw /paw/ paws ≻ pajs L W /saw/ sajs ≻ saws W L

• Grammar with lexical information
• And as more items are added...

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• Standard tableaux:
• Comparative tableau:
• Install *[σw
• Clone IDENT
• Grammar with lexical information
• *[σw ≫ IDENTpaw ≫ *ws ≫ IDENTsaw
• How does the grammar treat [paw] and [saw]?
• What are the predictions for a [w]-final word?
• And as more items are added...

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

12 / 18

• UR of plural suffix: /is/ or /s/
• With [w], [is] is out
• Standard tableaux:
• Comparative tableau:
• Install *[σw
• Clone IDENT
• Grammar with lexical information
• And as more items are added...
• *[σw ≫ IDENTpaw, gow, ZEw ≫

*ws ≫ IDENTsaw, anEw, bahiw, koketEw, fuÙibow, muzew, awkow

• The grammar of individual speakers may vary.

On average, the grammar will match the community’s lexicon.

• What are the predictions for a [w]-final word?
• Do the predictions differ for monosyllables and polysyllables?

Positional faithfulness

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

13 / 18

• Monosyllables are protected by initial syllable faithfulness
• Ranking arguments distinguish monosyllables from polysyllables
• A grammar with two lexical trends:
• Treatment of nonce words:
• Summary

Positional faithfulness

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

13 / 18

• Monosyllables are protected by initial syllable faithfulness
• (Trubetzkoy 1939; Steriade 1994; Beckman 1997, 1998;

Casali 1998; Barnes 2006; Jesney 2009; Becker 2009)

• [anEw ∼ anEjs] violates IDENT(back)
• [saw ∼ sajs] violates IDENT(back) and IDENT(back)σ

/ anEw + (i)s / IDENTσ *ws IDENT a. anEws *!

• b. ☞ anEjs

* / saw + (i)s / *ws IDENTσ IDENT

• a. ☞ sajs

* * b. saws *!

• Ranking arguments distinguish monosyllables from polysyllables
• A grammar with two lexical trends:
• Treatment of nonce words:

Positional faithfulness

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

13 / 18

• Monosyllables are protected by initial syllable faithfulness
• Ranking arguments distinguish monosyllables from polysyllables

IDENTσ IDENT *ws

/paw/ paws ≻ pajs

W W L

/saw/ sajs ≻ saws

L L W

/muzew/ muzews ≻ muzejs

W L

/anEw/ anEjs ≻ anEws

L W Clone in two stages: first IDENTσ, then IDENT: IDENTσpaw ≫ IDENTmuzew ≫ *ws ≫ IDENTσsaw, IDENTanEw

• A grammar with two lexical trends:
• Treatment of nonce words:
• Summary

Positional faithfulness

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

13 / 18

• Monosyllables are protected by initial syllable faithfulness
• Ranking arguments distinguish monosyllables from polysyllables
• A grammar with two lexical trends:

IDENTσpaw, gow, ZEw ≫ IDENTmuzew, kakaw

≫ *ws ≫ IDENTσsaw, mEw,

IDENTanEw, anzOw, bahiw, Zohnaw, hEpÙiw, posivew, tunew, nivew

• Monos have different ranking arguments from polys

→ monosyllables are listed separately from polysyllables

• Most monosyllables listed above *ws (F ≫ M)
• Most polysyllables listed below *ws (M ≫ F)
• Treatment of nonce words:
• Summary

Positional faithfulness

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

13 / 18

• Monosyllables are protected by initial syllable faithfulness
• Ranking arguments distinguish monosyllables from polysyllables
• A grammar with two lexical trends:
• Treatment of nonce words:
• Polysyllable: only IDENT and *ws matter:

/biñaw+(i)s/ IDσ60% ID20% *ws IDσ40% ID80% 20% biñaws * 80% biñajs (*) (*)

• Monosyllable: IDENTσ matters too:

/daw +(i)s/ IDσ60% ID20% *ws IDσ40% ID80% 60% daws * 40% dajs (*) (*) (*) (*)

• Summary

Positional faithfulness

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

13 / 18

• Monosyllables are protected by initial syllable faithfulness
• Ranking arguments distinguish monosyllables from polysyllables
• A grammar with two lexical trends:
• Treatment of nonce words:
• Summary
• Lexical items are separated into groups in response to

conflicting ranking arguments.

• Real lexical items are associated with clones, and thus

subject to a consistent ranking.

• Nonce words are more likely to be associated with the clone

that has more lexical items.

• What is the theoretical treatment of inter-speaker variation?
• What is the theoretical treatment of intra-speaker variation?

Positional surfeit

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

14 / 18

• Analysis of english plurals
• English Grammar
• Treatment of nonce words:
• Summary

Positional surfeit

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

14 / 18

• Analysis of english plurals

/ v@ômuT + z / IDENTσ *Ts IDENT

• a. ☞ v@ômuðz

* b. v@ômuTs *! / oUT + z / *Ts IDENTσ IDENT

• a. ☞ oUðz

* * b.

• UTs

*! Are *Ts and *fs reasonable constraints? What do we do with verbs, e.g. [SElf ∼ SElv]? Final voicing...? (Blevins 2004; Yu 2004; Kiparsky 2006; et seq.)

• English Grammar
• Treatment of nonce words:
• Summary

Positional surfeit

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

14 / 18

• Analysis of english plurals
• English Grammar

IDENTσstIf, brif ≫ IDENTheNk@ôÙIf, k@ôæf, irm2f, foUR@gôæf ≫ *Ts ≫ IDENTσnaIf, loUf, Ùif, ruf, oUT, Elf, wUlf, IDENTÃ@ôæf

• Monos have different ranking arguments from polys

→ monosyllables are listed separately from polysyllables

• Most monosyllables listed below *Ts (M ≫ F)
• Most polysyllables listed above *Ts (F ≫ W)
• Treatment of nonce words:
• Summary

Positional surfeit

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

14 / 18

• Analysis of english plurals
• English Grammar
• Treatment of nonce words:
• Polysyllable: only IDENT and *fs matter:

/m@leIf+z/ IDσ20% ID80% *fs IDσ80% ID20% 80% m@leIfs * 20% m@leIvz (*) (*)

• Monosyllable: IDENTσ too weak to matter:

/meIf+z/ IDσ20% ID80% *fs IDσ80% ID20% 80% meIfs * 20% meIvz (*) (*) (*) (*)

• Summary

Positional surfeit

• Overview

OT Basics Rankings Trends with constraint cloning

• Portuguese plurals
• Positional faithfulness
• Positional surfeit

Grammatical vs. respresentational approaches

14 / 18

• Analysis of english plurals
• English Grammar
• Treatment of nonce words:
• Summary
• In English, the high-ranking clone of general IDENT is strong,

so both monosyllables and polysyllables are protected.

• The grammar allows each individual existing word to behave

appropriately, but the generalization that monosyllables alternate more is not captured.

Grammatical vs. respresentational approaches

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

15 / 18

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

16 / 18

• If final [w] leads to assuming final /w/:
• If final [w] leads to different consonants in the UR:
• Grammatical vs. representational:

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

16 / 18

• If final [w] leads to assuming final /w/:

/ paw + (i)s / IDENTσ *ws

• a. ☞ paws

* b. pajs *! / saw + (i)s / *ws IDENTσ a. saws *!

• b. ☞ sajs

* Surface-true UR’s cause the grammar to be inconsistent

→ lexical information (and statistics) are in the grammar.

This is the grammatical approach.

• If final [w] leads to different consonants in the UR:
• Grammatical vs. representational:

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

16 / 18

• If final [w] leads to assuming final /w/:
• If final [w] leads to different consonants in the UR:

/ paw + (i)s / IDENTσ(back) *ws

• a. ☞ paws

* b. pajs *! / saL + (i)s / IDENTσ(back) *ws a. saws *!

• b. ☞ sajs

Abstract UR’s allow the grammar to be consistent

→ lexical information (and statistics) are only in the lexicon.

This is the representational approach.

• Grammatical vs. representational:

Portuguese plurals

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

16 / 18

• If final [w] leads to assuming final /w/:
• If final [w] leads to different consonants in the UR:
• Grammatical vs. representational:
• Representational:
• Keeps lexical trends in the lexicon; consistent grammar.

→ the analysis doesn’t capture the trends.

• How can these abstract UR’s be learned? Nobody knows.
• Grammatical:
• Moves lexical trends into the grammar(s)

→ grammatical effects are accounted for.

• No need to learn UR’s; UR = surface form of the base.

“inside-out” analysis (Hayes 1995, 1999).

• Need to learn and apply inconsistent grammars = easy.

OT vs. rules?

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

17 / 18

• Grammatical vs. representational = OT vs. rules
• But Optimality Theory is less dependent on representations
• Optimality Theory favors surface-true UR’s
• The grammatical approach encourages discovery of

generalizations

OT vs. rules?

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

17 / 18

• Grammatical vs. representational = OT vs. rules
• Grammatical rule-based analysis is possible
• Lexically-specific rules are in SPE (Chomsky & Halle 1968)
• Not popular because Halle likes consistent grammars
• But Optimality Theory is less dependent on representations
• Optimality Theory favors surface-true UR’s
• The grammatical approach encourages discovery of

generalizations

OT vs. rules?

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

17 / 18

• Grammatical vs. representational = OT vs. rules
• But Optimality Theory is less dependent on representations
• *NC

˚ (Pater 1999)

• Phonological processes don’t have to be assimilation,

dissimilation, or simplification.

• Optimality Theory favors surface-true UR’s
• The grammatical approach encourages discovery of

generalizations

OT vs. rules?

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

17 / 18

• Grammatical vs. representational = OT vs. rules
• But Optimality Theory is less dependent on representations
• Optimality Theory favors surface-true UR’s
• “Lexicon Optimization” (=assume surface-true UR unless

there is evidence to the contrary)

• The grammatical approach encourages discovery of

generalizations

OT vs. rules?

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

17 / 18

• Grammatical vs. representational = OT vs. rules
• But Optimality Theory is less dependent on representations
• Optimality Theory favors surface-true UR’s
• The grammatical approach encourages discovery of

generalizations

• The grammatical approach works at the morpheme level;

The representational approach works at the segment level.

• In the morpheme-level analysis, the alternation has to be

localized by the grammar: Russian: [vetir ∼ vetr-a], but not *[veter ∼ vtera]

References

• Overview

OT Basics Rankings Trends with constraint cloning Grammatical vs. respresentational approaches

• Portuguese plurals
• OT vs. rules?
• References

18 / 18

Barnes, Jonathan (2006). Strength and Weakness at the Interface: Positional Neutralization in Phonetics and Phonology. Berlin/New York: Mouton de Gruyter. Becker, Michael (2009). Phonological Trends in the Lexicon: The Role of Constraints. Ph.D. dissertation, University of Massachusetts Amherst. Beckman, Jill (1997). Positional faithfulness, positional neutralisation and Shona vowel harmony. Phonology 14. 1–46. Beckman, Jill (1998). Positional Faithfulness. Ph.D. dissertation, University of Massachusetts Amherst, Amherst, MA. Blevins, Juliette (2004). Evolutionary phonology: the emergence of sound patterns. Cambridge: Cambridge University Press. Casali, Roderic (1998). Resolving Hiatus. Garland, New York. Chomsky, Noam & Morris Halle (1968). The Sound Pattern of