Irreducible Parallelism and Desirable Serialism John McCarthy - - PowerPoint PPT Presentation

1 Irreducible Parallelism and Desirable Serialism John McCarthy UMass Amherst

2 Background  Harmonic Serialism (HS) is a version of OT with a Gen → Eval → Gen → … loop (Prince & Smolensky, McCarthy).  HS’s Gen, unlike parallel OT’s, is limited to making one change at a time.  Candidates derived by all single changes compete.  Winner goes back into Gen.  This process continues until convergence, when output = input.  E.g., Gen(ktub) = {ktub, uktub, kutub, tub, ktu, …} Gen(uktub) = {uktub, ʔuktub, ktub, uktu, …} Gen(ʔuktub) = {ʔuktub, uktub, ʔuktu, …}  Architectural imperative of HS: harmony improves monotonically through derivation (relative to H).

3 Research Problem and (Ultimate) Goal  Properties of Gen in parallel OT are not usually deemed to be

• f much interest, except for correspondence theory and

representational universals (e.g., Ft > σ in prosodic hierarchy).  Not so in HS. Much depends on what “one change” is — in

• ther words, how is Gen defined.

 In today’s talk, I will:

• Explain why the “one change” question is important, and

how it can be studied.

• Scrutinize some particularly challenging cases where it

looks like one change is really two (“irreducible parallelism”).  The discussion will rely on certain recurrent themes of HS:

4 Recurrent Themes  No look-ahead: Selection of intermediate optima can’t refer to potential for later improvements. Decisions are made locally, based on information available at that point in the derivation.  Continuous availability of operations: Gen’s operations are not limited to specific derivational steps. They are always freely available.  Emergence of temporary ill-formedness: Violations of markedness constraints, even surface-true ones, may be introduced in the course of the derivation and later eliminated.  Corollary: Apparent non-monotonicity of harmonic improvement: HS architecture guarantees monotonic harmonic improvement through derivation, but it might not always look that way.  Corollary: Revelation of ill-formedness. A structure’s ill- formedness may be disclosed in the course of a derivation, as

• ther structure is built

5 Stating the Question  Assume Gen has certain primitive operations: insert, delete, associate, … (perhaps like Archangeli & Pulleyblank’s parametric rule theory).  Simplest hypothesis: “one change” = a single application of

• ne primitive operation.
• Gen(x) = {x, op1(x), op2(x), …} (opn(x) = result of

applying the primitive operation opn at some locus in x.)  Question: Does simplest hypothesis suffice? Does Gen ever include candidates like this?

• Gen(x) = {…, opk(opj(x)), …}

 Question: If simplest hypothesis doesn’t suffice, are there principled limits on combining primitive operations?  Terminology: “Parallel” even if sequential Gen-internally.

6 Irreducible Parallelism  How can we infer that certain primitive operations must be allowed to apply in parallel?  More concretely:

• I → O mapping is observed, where O = op2(op1(I)).
• Is M = op1(I) a necessary intermediate step I → M → O?
• Or must op1 and op2 apply in parallel, skipping M?

 A priori, I → M → O is preferable because it implies a simpler Gen.  But I → O may be unavoidable if rankings required for I → M and M → O mappings are inconsistent with each other, rest

• f language, or UG. In that case, op1 and op2 are irreducibly
• parallel. (Putative examples later.)

 Irreducible parallelism responds to the threat of typological undergeneration: I → O mapping needs an analysis.

7 Desirable Serialism  Arguments for HS over parallel OT are based on avoiding typological overgeneration:

• I → O mapping is never observed, where O = op2(op1(I)).
• Suppose ranking permutation predicts this mapping in

parallel OT. This is typological overgeneration.  Desirable serialism:

• Suppose HS mandates intermediate M = op1(I) because
• p1 and op2 are not allowed to apply in parallel in Gen.

 If rankings required for I → M and M → O mappings are inconsistent with each other, rest of language, or UG, then HS correctly predicts impossibility of unobserved I → O mapping.  This is desirable serialism: desirable because it avoids typological overgeneration.

8 Desirable Serialism Exemplified (Jesney)  Positional faithfulness wrongly predicts modification of position to facilitate neutralization (Noyer). E.g., initial stress except when first vowel is reducible and second isn’t: /bedu/ → ˈbedu /kaza/ → ˈkazə but /patu/ → pəˈtu IDˈσ(low) HEAD(Wd) *LOW ALIGN- L(ˈσ) ID(low) → ˈkazə 1 1 kəˈza 1 1 W 1 ˈkaza 2 W L ˈkəza 1 W 1 1 ˈkəzə 1 W L 2 W kəzə 1 W L 2 W → pəˈtu 1 1 ˈpatu 1 W L L ˈpətu 1 W L 1 pətu 1 W 1

9  Problem arises because stress assignment and vowel reduction can vary together across candidates. This is standard for parallel OT.  Now assume that stress assignment and vowel reduction are separate operations that cannot apply in parallel in HS. So candidates can vary in stress location or reduction, but not both: GenHS(/patu/) = {patu, ˈpatu, paˈtu, pətu}  Winner at Step 1 is ˈpatu: IDˈσ(low) HEAD(Wd) *LOW AL-L(ˈσ) ID(low) → ˈpatu 1 pətu 1 W L 1 W patu 1 W 1 paˈtu 1 1 W  Derivation converges on ˈpatu at Step 2. Problematic pəˈtu is never even a candidate.  This is desirable serialism: typological overgeneration is avoided if reduction and stress can’t apply in parallel.

10 Further Claims About Desirable Serialism  Disparate processes

• Stress and segmental processes (Jesney, Staubs)
• Syllabification and segmental processes (Jesney)
• Prosodic parsing and epenthesis (Moore-Cantwell)

 Same process (iteration, multiple application):

• Syllabification (Elfner, Pater)
• Metrical foot assignment (Pruitt)
• Deletion of segments and features (McCarthy)
• Metathesis (McCarthy)
• Autosegmental spreading (McCarthy)
• Epenthesis (Kimper)

 Derivational evidence

• Stress and syncope (McCarthy)
• Stress and epenthesis (Elfner)
• Mora insertion and segmental deletion (Torres-Tamarit)

11 The Challenge  HS research program depends on defining Gen thus:

• Gen must accommodate observed cases of irreducible

parallelism (to avoid undergeneration).

• Gen should accommodate claims about desirable serialism

(to avoid overgeneration).

• Departures from total serialism — that is, primitive
• perations that are irreducibly parallel — should be

principled — if they exist at all.  Contrived, expendient, or free-ranging parallelism threatens HS research program by undermining desirable serialism and ultimately blurring line between HS and parallel OT.  We’ll now examine some phenomena that look like irreducible parallelism but turn out not to be.

12 Cross-Level Interactions

13 Cross-Level Interactions  From McCarthy, Pater, & Pruitt (to appear).  Cross-level interaction (CLI): Markedness constraints on higher-level structure demand change in lower level:

• Latin “iambic” shortening: /amoː/ → (ˈamo), with

preference for LL over LH trochees forcing shortening and FOOT-BINARITY overriding extrametricality (Mester).

• Hixkaryana “unstressable word syndrome” (Hayes): /kana/

→ (ˈkaː)na, with HEAD(PrWd) and FT-BIN forcing

• therwise impossible (ˈCVː) foot.

 CLIs are evidence against a particular kind of serialism, called “bottom-up constructionism” (BUC) by P&S: derivations are strictly bottom up, with no revision of earlier structure:

• Underlying

amoː kana

• Shortening/lengthening

Not motivated

• Footing

(ˈamoː) (ˈka)na

• Shortening/lengthening

Not available

14 CLIs and Parallelism  Parallel OT avoids the BUC problem by evaluating effects of footing and shortening/lengthening together, in parallel: /kana/ HEAD(Wd) NON-FIN(ft) FT-BIN DEP(µ) Operations → (ˈkaː)na 1 Footing, lengthening kana 1 W L None (ˈkana) 1 W L Footing (ˈka)na 1 W L Footing kaːna 1 W 1 Lengthening  Under BUC, footing and shortening/lengthening have to be done in parallel because they can’t be done serially

15 CLIs and Harmonic Serialism  HS isn’t BUC. All single operations are available at every step

• f the derivation, top-down as well as bottom-up:

Step 1 HEAD(Wd) NON-FIN(ft) FT-BIN DEP(µ) Operations → (ˈka)na 1 W

Footing

kana 1 W

None

(ˈkana) 1 W

Footing

kaːna 1 W 1 W

Lengthening

Step 2 HEAD(Wd) NON-FIN(ft) FT-BIN DEP(µ) Operations → (ˈkaː)na 1

Lengthening

(ˈka)na 1 W L

None

(ˈka)(ˈna) 1 W 2 W L

Footing

Step 3 — convergence. Conclusion: Lengthening/shortening and footing are not irreducibly parallel.

16 Desirable Serialism  Hixkaryana does not use /…CV…/ → …(ˈCVː)… mapping to better satisfy PARSE-SYLLABLE:

• /aʧowowo) → (aˈʧoː)wowo, *(aˈʧoː)(ˈwoː)wo

 In HS analysis, this is straightforward: FT-BIN >> PARSE- SYLL, so (ˈwo) loses at intermediate step and (ˈwoː) is never even a candidate.  In parallel analysis, that solution isn’t available because (ˈwoː) satisfies FT-BIN. More ad hoc measures must be adopted, such as atypical constraint on iambs (Kager textbook).

17 Discussion  Irreducible parallelism -- No. Foot parsing and mora deletion/insertion are not irreducibly parallel. They can apply serially.  Desirable serialism. A case can be made that foot parsing and mora deletion/insertion not only aren’t necessarily parallel but should be serial.  Full availability of structural operations. HS is not BUC. Mora deletion/insertion operations remain available even after feet are built.  Emergence of temporary ill-formedness. Hixkaryana has no surface (ˈCV) feet, but they may emerge in the course of the derivation, only to be eliminated later. (Likewise Latin (ˈamoː).)

18  Apparent non-monotonicity of harmonic improvement. Harmony in fact improves monotonically, relative to the constraint hierarchy where HEAD(word) and NON- FINALITY(foot) dominate FOOT-BINARITY.  No look-ahead. Although the (ˈCV) foot is eventually brought into conformity with FOOT-BINARITY, look-ahead is

• unnecessary. The ranking FOOT-BINARITY >> DEP(µ) takes

care of things at Step 2.

19 Tongan  In Tongan, long vowels are parsed heterosyllabically to ensure right-aligned (ˈLL) trochee (Churchward, Poser):

• mo(ˈheŋa)

koto(ˈkoː) hu(ˈufi)  This is a CLI. In parallel OT, foot parsing and syllabification are optimized in parallel (P&S): /kotoko:/ ALIGN-R(ft) ONSET → ko.to.(ˈkoː) ko.to.(ˈko.o) 1 W /hu:fi/ ALIGN-R(ft) ONSET → hu.(ˈu.fi) 1 (ˈhu:).fi 1W L

20 HS Analysis (Elsman)  WEIGHT-TO-STRESS (WTS) forces V.V syllabification prior to stress assignment: Step 1 WTS ONSET → ko.to.ko.o 1 ko.to.koː 1 W L Step 1 WTS ONSET → hu.u.fi 1 huː.fi 1 W L  Stress is assigned at Step 2: ko.to.(ˈko.o), hu.(ˈu.fi)

21  At Step 3, (ˈko.o) can now fuse without violating WTS: Step 3 WTS ONSET → ko.to.(ˈkoː) 1 ko.to.(ˈko.o) L  No fusion for hu.(ˈu.fi) at Step 3, however:

• *(ˈhuː.fi) is out because of uneven trochee.
• *(ˈhuː.).fi isn’t even a candidate, because it has

restructured syllable and foot together, in parallel. Remark: Hard to know what to make of syllable fusion operation. Connected with claims about OCP (Levin [Blevins]). But point stands that syllabification and footing needn’t occur in parallel.

22 Discussion  Irreducible parallelism. Foot parsing and syllabification are not irreducibly parallel. This is consistent with results about foot parsing and mora operations.  Full availability of structural operations. Again, HS is not

• BUC. Syllabification can be changed even after feet are built.

 Emergence of temporary ill-formedness, apparent non- monotonicity of harmonic improvement. In the course of the derivation, koto(ˈkoː) violates ONSET only to obey it later. P&S note that Poser’s similar analysis must start with marked ONSET-violating syllable syllabification. Elsman’s insight is that it’s not marked, because WTS dominates ONSET.  Revelation of ill-formedness. Assignment of stress in effect reveals ko.to.(ˈko.o)’s ill-formedness, which necessitates syllable fusion.

23 Faithfulness Effects on Structure

24 Metrical Parsing in Shipibo  Disyllabic feet left to right.  Underlying weight of odd-even pair determines trochee/iamb choice:

• Trochee:
• LL → (ˈLL): /CVCVCVCV…/ → (ˈCVCV)(ˈCVCV)…
• HL → (ˈLL): /CVCVCVːCV…/ → (ˈCVCV)(ˈCVCV)…
• Iamb:
• LH → (LˈH): /CVCVCVCVː…/ → (ˈCVCV)(CVˈCVː)…
• HH → (LˈH): /CVCVCVːCVː…/ → (ˈCVCV)(CVˈCVː)…

 Thanks to Minta Elsman for pointing out relevance of

Shipibo.

25 Parallel OT Analysis (Ulloa)  GROUPING HARMONY (GH – Prince) favors (ˈLL) trochees and

(LˈH) iambs.

 Trochaic rhythm is the default (TROCH >> IAMB).  Faithfulness favors iambic parse of LH and HH because MAX(µ) >> TROCH:

• LH → (LˈH), *(ˈLL) avoids any MAX(µ) violation
• HH → (LˈH), *(ˈLL) avoids one MAX(µ) violation.

 /…LL/ case: …CVCV GH MAX(µ) TROCH IAMB → ...(ˈCVCV) 1 ...(CVˈCV) 1 W L

26  /…HL/ case: …CVːCV GH MAX(µ) TROCH IAMB → ...(ˈCVCV) 1 1 ...(ˈCVːCV) 1 W L 1  /…LH/ case: …CVCVː GH MAX(µ) TROCH IAMB → ...(CVˈCVː) 1 ...(ˈCVCV) 1 W L 1 W ...(ˈCVCVː) 1 W L 1 W

27  /…HH/ case: …CVːCVː GH MAX(µ) TROCH IAMB → ...(CVˈCVː) 1 1 ...(ˈCVCV) 2 W L 1 W ...(CVːˈCVː) 1 W L 1 ...(ˈCVːCVː) 1 W L L 1 W

28 Discussion  Ulloa’s analysis depends on evaluating results of foot parsing and vowel shortening together, in parallel.  Irreducible parallelism? It could not be reproduced in HS unless parsing and shortening operations were irreducibly parallel.  General result of no look-ahead in HS:

• Only the immediate faithfulness consequences of an
• peration can affect its applicability in winning candidate.
• Thus, MAX(µ) can affect shortening, but not foot parsing.

 Pruitt: This is desirable serialism. It rules out non-local interactions between foot parsing and shortening. E.g.,

• L→R trochees: (ˈLL)(ˈLL)L.
• HLeven no shortening: (ˈH)(ˈLL)(ˈLL)
• HLodd shortens for PARSE-SYLL: (ˈLL)(ˈLL) ≻ (ˈH)(ˈLL)L

29 HS Analysis  Trochee/iamb choice can be determined by markedness, pre- shortening --- faithfulness isn’t the only option.

• (XH) parsed iambically.
• (XL) parsed trochaically.

 Assume some markedness constraint “*(ˈXH)”.  Then interesting derivations proceed as follows:  /…HL/ case (after parsing initial CVCV at Step 1): Step 2 PARSE-SYLL *(ˈXH) GH MAX(µ) TROCH IAMB → ...(ˈCVːCV) 1 1 ...(CVːˈCV) 1 1 W L ...CVːCV 2 W L L ...CVCV 2 W L 1 W L

30 Step 3 PARSE-SYLL *(ˈXH) GH MAX(µ) TROCH IAMB → ...(ˈCVCV) 1 1 ...(ˈCVːCV) 1 W L 1  /…HH/ case: Step 2 PARSE-SYLL *(ˈXH) GH MAX(µ) TROCH IAMB → ...(CVːˈCVː) 1 1 ...(ˈCVːCVː) 1 W 1 L 1 W ...CVːCV 2 W L 1 W L Step 3 PARSE-SYLL *(ˈXH) GH MAX(µ) TROCH IAMB → ...(CVˈCVː) 1 1 ...(CVːˈCVː) 1 W L 1

31 Further Detail  CVC syllables are heavy exactly where CVː is permitted (i.e., stressed position of iambic foot).  Elsman: This seems to require irreducible parallelism of foot parsing and insertion of weight-by-position moras. More generally, must foot parsing and foot-form optimization occur in parallel?  Serial alternative: WEIGHT-BY-POSITION >> PARSE-

• SYLLABLE. At point of foot parsing, all CVC are heavy. Same

ranking that produces shortening of CVː also produces de- weighting of CVC. (MAX(µ) violated in both cases.).  This is revelation of ill-formedness — violations of GH emerge when feet are created.

32 Resyllabification

33 Resyllabification  Claim (McCarthy): Resyllabification has to be able to apply in parallel with other operations.  Argument comes from syncope in “two-sided open syllable” — i.e., VC___CV context.

34 Syncope in Two-Sided Open Syllable  Syncope cannot create CCC clusters:

• /patika/ → pat.ka
• /paktika/ → paktika

/pakitka/ → pakitka  Standard analysis: syncope is blocked from creating unsyllabifiable consonants: /patika/ *COMPLEX PARSE-SEG “SYNCOPE” MAX → pat.ka 1 pa.ti.ka 1 W L /paktika/ *COMPLEX PARSE-SEG “SYNCOPE” MAX → pak.ti.ka 1 pakt.ka 1 W L 1 W pak.t.ka 1 W L 1 W

35  Irreducible parallelism? This analysis requires evaluating the consequences of syncope and resyllabification together, in parallel.  If syncope and syllabification do not apply in parallel, derivation must proceed …→ pa.ti.ka → pa.t.ka → pat.ka.  But then why not …→ pak.ti.ka → *pak.t.ka?  No look-ahead: HS can’t look ahead to impossibility of resyllabifying *pak.t.ka.  So derivation must proceed …→ pa.ti.ka → pat.ka, with syncope and syllabification operations applying in parallel.  Rationale offered (McCarthy):

• Syllabification is non-contrastive.
• Hence, (re)syllabification doesn’t violate faithfulness.
• Processes that don’t violate faithfulness are “free” – they

can apply in parallel with processes that do (e.g. syncope).

36 Discussion  Claim about syllabification is problematic:

• There are arguments that syllabification can’t even apply

in parallel with itself (Elfner, Pater).

• Elsman: syllabification and footing aren’t irreducibly

parallel in Tongan.

• Desirable serialism: HS solution to positional faithfulness

problem requires serial syllabification and segmental processes:  Parallel OT makes wrong prediction (Noyer) /bada/ PARSE- SEG IDOns(voice) *VOICE ID(voice) ONSET → bat.a 1 1 1 ba.da 2 W L L ba.ta 1 W 1 1 L pat.a 1 W L 2 W 1 pata 4 W L 2 W

37  HS makes correct prediction (Jesney), if syllabification and devoicing aren’t parallel: Step 1 PARSE- SEG IDOns(voice) *VOICE ID(voice) ONSET → ba.da 2 bata 4 W 1 L 1 W pada 4 W 1 L 1 W bad.a 2 1 W Step 2

(convergence)

PARSE- SEG IDOns(voice) *VOICE ID(voice) ONSET → ba.da 2 ba.ta 1 W 1 L 1 W

38 Discussion (continued)  Claim about rationale is also problematic:

• Pruitt: Footing doesn’t violate faithfulness, but it can’t

apply in parallel with itself or other processes.

• Faithfulness to derived syllabification.
• Elfner: Contrastive syllabification in Blackfoot.

 There are other ways of analyzing two-sided open syllable syncope.

39 Some Alternatives  Non-syllable-based phonotactics (Steriade):

• In patka, t is licensed by VC transition cues.
• In paktka, t is unlicensed by VC transition cues.

/patika/ LIC-BY-CUE “SYNCOPE” MAX → patka 1 patika 1 W L /paktika/ LIC-BY-CUE “SYNCOPE” MAX → paktika 1 paktka 1 W L 1 W

• Resyllabification is irrelevant.

40  Onset/stress relation:

• Aranda: stress skips initial onsetless syllable.
• English (Kahn, Selkirk): onsets are attracted out of medial

unstressed syllables (as shown by flapping): caɾ.a.pult, cap.iɾ.al.

• If weak syllables avoid onsets, and if syncope is a

response to weakness, then derivation could be: pa.ti.ka → pat.i.ka → pat.ka with harmonically improving resyllabification before syncope.  Again, there’s no compelling case for irreducible parallelism.

41 Discussion  No look-ahead in HS seems to cause difficulties for analysis

• f two-sided open syllable syncope: decision about whether to

syncopate depends on knowing consequences of resyllabification.  This observation led McCarthy to propose that faithful

• perations are irreducibly parallel with unfaithful ones.

 But this claim is problematic, undermining cases of desirable serialism.  It is moreover unnecessary: there are other ways of analyzing two-sided open syllable syncope that do not require resyllabification.  The Licensing-by-Cue alternative suggests a general analytic strategy: if full sequelae of an operation aren’t immediately known but are relevant to evaluation, look for a proxy — e.g., string-based rather than structure-based phonotactics.

42 Iterative Processes

43 Process Iteration in HS  In rule-based phonology:

• Simultaneous application (SPE, Anderson): Process

applies to all eligible loci at once, in parallel.

• Iterative application (Howard, Johnson, …): Process

applies to some locus, then another, serially.  Parallel OT has simultaneous application: all feet parsed at

• nce; all spreading targets affected at once; etc.

 In HS, Gen can be defined to require iterative application: feet parsed one at a time; spreading affects one segment/syllable at a time; etc.

44 Desirable Serialism  Pruitt: Iterative foot parsing avoids unwanted prediction of non-local effects on quantity (mentioned previously).  McCarthy, Wilson: Iterative spreading accounts for myopia in harmony.  Myopia: Feature will spread toward a target (e.g., peripheral

• r stressed syllable) even if blocked from reaching it.

Blocking never entails complete failure (‘no sour grapes’).  Myopia follows from no look-ahead property of HS, if spreading is iterative. Iterative local spreading toward non- local target must be done with gradient constraints (e.g., ALIGN), which favor partial spreading.

45 Non-Myopic Spreading in Central Veneto  Walker:

• High suffix raises preceding ˈe, ˈo: kalˈseto/kalˈsiti,

ˈmovo/ˈmuvi.

• In proparoxytones, unstressed penult e, o affected too:

/ˈorden-i/ → ˈurdini.

• But not if antepenult isn’t a legitimate target: /ˈangol-i/ →

ˈangoli, *ˈanguli.

• Spreading also blocked by non-targetable penult: /laˈvorav-

i/ → laˈvoravi, *laˈvuravi.  Conclusions:

• Harmony is not myopic.
• Effects of spreading to both syllables in /ˈorden-i/ and

/ˈangol-i/ need to be evaluated together, in parallel.

• Tableaux on next slide:

46 /ˈorden-i/ ID(low) LOCALITY LIC-HIGH ID(high) → ˈurdini 2 ˈurdeni 1 W 1 L ˈordeni 1 W L /ˈangol-i/ ID(low) LOCALITY LIC-HIGH ID(high) → ˈangoli 1 ˈanguli 1 1 W ˈinguli 1 W L 2 W /laˈvorav-i/ ID(low) LOCALITY LIC-HIGH ID(high) → laˈvoravi 1 laˈvuravi 1 W L 1 W

47 Reanalysis (Kimper)  Emergence of temporary ill-formedness: HS derivation is /ˈorden-i/ → ˈurdeni → ˈurdini, with skipping at intermediate

• step. No-skipping requirement is violated in course of

derivation, though obeyed in surface forms: Step 1 ID(low) *SKIP(low) LIC-HIGH *SKIP(mid) ID(high) → ˈurdeni 1 1 ˈordeni 1 W L L Step 2 ID(low) *SKIP(low) LIC-HIGH *SKIP(mid) ID(high) → ˈurdini 1 ˈurdeni 1 W L

48 Step 1 ID(low) *SKIP(low) LIC-HIGH *SKIP(mid) ID(high) → ˈangoli 1 ˈanguli 1 1 W ˈinguli 1 W L 1 W Step 1 ID(low) *SKIP(low) LIC-HIGH *SKIP(mid) ID(high) → laˈvoravi 1 laˈvuravi 1 W L 1 W laˈvorivi 1 W 1 1 W

49 Substantive Difference  Parallel analysis: laˈvoravi, *laˈvuravi because a is not a target

• f harmony (high-ranking IDENT(low)).

 Serial analysis: laˈvoravi, *laˈvuravi because a is not skippable (high-ranking *SKIP(low)).  Parallel analysis has undifferentiated LOCALITY.  Serial analysis needs differentiated *SKIP(low) and *SKIP(mid).  Kimper: typological evidence supports the differentiated no- skipping constraints.

50 Reduplicative Overapplication

51  Overapplication: Phonological process applies in both reduplicant and base, though structural description is met in

• nly one of them: Chumash /k-RED-ʔaniʃ/ → kˀan-kˀaniʃ.

 Most can be analyzed serially (Wilbur, Kiparsky, etc.), but a few seem to require parallelism: Malay /RED-waŋi/ → w ŋ - w ŋ (Onn, Kenstowicz, McCarthy & Prince).  The Malay example has been questioned on empirical grounds (Inkelas & Zoll, Kiparsky), and the handful of other cases (Axininca, Paamese, and Korean) can be reanalyzed (McCarthy, Kimper, & Mullin).  Reduplication/phonology interactions do not present a compelling case of irreducible parallelism.

52 Summary  How serial is Harmonic Serialism?  Evidence from desirable serialism.  Reanalysis of cases of irreducible parallelism.  Foundation has been laid for more intensive investigation of Gen’s set of operations.

53

References Anderson, Stephen R. (1974) The Organization of Phonology. New York: Academic Press. Chomsky, Noam and Morris Halle (1968) The Sound Pattern of

• English. New York: Harper & Row.

Elfner, Emily (2006) Contrastive syllabification in Blackfoot. In Donald Baumer, David Montero and Michael Scanlon (eds.), Proceedings of the 25th West Coast Conference on Formal Linguistics, 141-149. Somerville, MA: Cascadilla Press. Elfner, Emily. (2007) Moraic faithfulness: Evidence from Blackfoot and

• English. Unpublished paper, University of Massachusetts Amherst,

Amherst, MA. Handout of paper presented at the 15th Manchester Phonology Meeting, University of Manchester, Manchester, England, May 26. Available at http://www.people.umass.edu/eelfner/morafaith.pdf. Elfner, Emily (to appear) Stress-epenthesis interactions in Harmonic

• Serialism. In John J. McCarthy and Joe Pater (eds.), Harmonic

Grammar and Harmonic Serialism. London: Equinox Publishing.

54

Elias-Ulloa, Jose (2005) Rhythmic consequences of Panoan metrical

• phonology. Doctoral dissertation, Rutgers University, New

Brunswick, NJ. Elsman, Minta (to appear) Parallelism vs. serialism, or constraints vs. rules? Tongan stress and syllabification revisited. In John J. McCarthy and Joe Pater (eds.), Harmonic Grammar and Harmonic

• Serialism. London: Equinox Publishing.

Howard, Irwin (1972) A Directional Theory of Rule Application in

• Phonology. Doctoral dissertation, MIT, Cambridge, MA.

Inkelas, Sharon and Cheryl Zoll (2005) Reduplication: Doubling in

• Morphology. Cambridge ; New York: Cambridge University Press.

Jesney, Karen (to appear) Positional faithfulness, non-locality, and the Harmonic Serialism solution. In Suzi Lima, Kevin Mullin and Brian Smith (eds.), Proceedings of the 39th Annual Meeting of the North East Linguistic Society. Amherst, MA: GLSA Publications. Available on Rutgers Optimality Archive, ROA-1018. Johnson, C. Douglas (1971) Unbounded expressions in rules of stress and accent. Glossa 4: 185-196.

55

Johnson, C. Douglas (1972) Formal Aspects of Phonological

• Description. The Hague: Mouton.

Kahn, Daniel (1976) Syllable-based Generalizations in English

• Phonology. Doctoral dissertation, MIT, Cambridge, MA. Published

by Garland Press, New York, 1980. Kenstowicz, Michael (1981) Functional explanations in generative

• phonology. In D. L. Goyvaerts (ed.), Phonology in the 1980's, 431-
• 444. Ghent: E. Story-Scientia.

Kimper, Wendell (2011) Locality and globality in phonological

• variation. Natural Language & Linguistic Theory 29: 423-465.

Kimper, Wendell (2012) Harmony is myopic. Linguistic Inquiry 43: 301-309. Kimper, Wendell (to appear) Positive constraints and finite goodness in Optimality Theory. In John J. McCarthy and Joe Pater (eds.), Harmonic Grammar and Harmonic Serialism. London: Equinox Publishing. Kiparsky, Paul (2010) Reduplication in Stratal OT. In Linda Uyechi and Lian Hee Wee (eds.), Reality Exploration and Discovery: Pattern

56

Interaction in Language & Life, 125-142. Stanford, CA: CSLI Publications. McCarthy, John J. (2000) Harmonic serialism and parallelism. In Masako Hirotani (ed.), Proceedings of the North East Linguistics Society 30, 501-524. Amherst, MA: GLSA Publications. Available

• n Rutgers Optimality Archive, ROA-357.

McCarthy, John J. (2007) Restraint of analysis. In Sylvia Blaho, Patrik Bye and Martin Krämer (eds.), Freedom of Analysis, 203-231. Berlin and New York: Mouton de Gruyter. Also appears in Eric Bakovic, Junko Ito, and John J. McCarthy, eds. (2006) Wondering at the Natural Fecundity of Things: Essays in Honor of Alan Prince. Linguistics Research Center, UC Santa Cruz. Available at http://repositories.cdlib.org/lrc/prince/10. McCarthy, John J. (2008a) The gradual path to cluster simplification. Phonology 25: 271-319. doi:10.1017/S0952675708001486. McCarthy, John J. (2008b) Harmony in Harmonic Serialism. Unpublished paper, University of Massachusetts Amherst, Amherst,

• MA. Available on Rutgers Optimality Archive, ROA-1009.

57

McCarthy, John J. (2008c) The serial interaction of stress and syncope. Natural Language & Linguistic Theory 26: 499-546. DOI: 10.1007/s11049-008-9051-3. McCarthy, John J. (2009) Studying Gen. Journal of the Phonetic Society of Japan 13: 3-12. McCarthy, John J. (2010) An introduction to Harmonic Serialism. Language and Linguistics Compass 4: 1001-1018. McCarthy, John J. (2011) Perceptually grounded faithfulness in Harmonic Serialism. Linguistic Inquiry 42. Available at http://works.bepress.com/john_j_mccarthy/104/. McCarthy, John J. (to appear) The theory and practice of Harmonic

• Serialism. In John J. McCarthy and Joe Pater (eds.), Harmonic

Grammar and Harmonic Serialism. London: Equinox Publishing. McCarthy, John J., Wendell Kimper and Kevin Mullin (2012) Reduplication in Harmonic Serialism. Morphology 22: 173-232. McCarthy, John J., Kevin Mullin and Brian W. Smith (2012) Implications of Harmonic Serialism for lexical tone association. In

• E. D. (Bert) Botma and Roland Noske (eds.), Phonological

58

Architecture: Empirical, Theoretical and Conceptual Issues - Papers in Honour of Norval S.H. Smith. . Berlin: Mouton de Gruyter. McCarthy, John J., Joe Pater and Kathryn Pruitt (to appear) Cross-level interactions in Harmonic Serialism. In John J. McCarthy and Joe Pater (eds.), Harmonic Grammar and Harmonic Serialism. London: Equinox Publishing. McCarthy, John J. and Kathryn Pruitt (2012) Sources of phonological

• structure. In Ralf Vogel and Hans Broekhuis (eds.), Linguistic

Derivations and Filtering: Minimalism and Optimality Theory. London: Equinox Publishing. Moore-Cantwell, Claire (to appear) Contexts for epenthesis in Harmonic Serialism. In John J. McCarthy and Joe Pater (eds.), Harmonic Grammar and Harmonic Serialism. London: Equinox Publishing. Onn, Farid M. (1980) Aspects of Malay Phonology and Morphology: A Generative Approach. Kuala Lumpur: Universiti Kebangsaan Malaysia. Pater, Joe (2012) Serial Harmonic Grammar and Berber syllabification. In Toni Borowsky, Shigeto Kawahara, Takahito Shinya and Mariko

59

Sugahara (eds.), Prosody Matters: Essays in Honor of Lisa Selkirk. London: Equinox Publishing. Pater, Joe and Robert Staubs (to appear) Learning serial constraint- based grammars. In John J. McCarthy and Joe Pater (eds.), Harmonic Grammar and Harmonic Serialism. London: Equinox Publishing. Poser, William J. (1985) Cliticization to NP and Lexical Phonology. The Proceedings of the West Coast Conference on Formal Linguistics 4, 262-272. Prince, Alan and Paul Smolensky (1993/2004) Optimality Theory: Constraint Interaction in Generative Grammar. Malden, MA, and Oxford, UK: Blackwell. Pruitt, Kathryn (2010) Serialism and locality in constraint-based metrical parsing. Phonology 27: 481-526. Pruitt, Kathryn (2012) Stress in Harmonic Serialism. Doctoral dissertation, University of Massachusetts, Amherst, Amherst, MA. Selkirk, Elisabeth (1982) The syllable. In Harry van der Hulst and Norval Smith (eds.), The Structure of Phonological Representations, 337-383. Dordrecht: Foris.

60

Staubs, Robert (to appear) Serial restrictions on the interactions of features and stress. In John J. McCarthy and Joe Pater (eds.), Harmonic Grammar and Harmonic Serialism. London: Equinox Publishing. Torres-Tamarit, Francesc Josep (to appear) Compensatory and vowel lengthening in Harmonic Serialism. In John J. McCarthy and Joe Pater (eds.), Harmonic Grammar and Harmonic Serialism. London: Equinox Publishing. Walker, Rachel (2010) Nonmyopic harmony and the nature of

• derivations. Linguistic Inquiry 41: 169-179.

Wilson, Colin. (2003) Unbounded spreading in OT (or, Unbounded spreading is local spreading iterated unboundedly). Unpublished paper, UCLA. Wilson, Colin. (2004) Analyzing unbounded spreading with constraints: Marks, targets, and derivations. Unpublished paper, UCLA. Wilson, Colin. (2006) Unbounded spreading is myopic. Unpublished paper, UCLA. Available at http://www.linguistics.ucla.edu/people/wilson/Myopia2006.pdf.