[Author Login]
[Home]
ROA:282
Title:Positional Asymmetries and Licensing
Authors:Cheryl Zoll
Comment:28 pages. Word and RTF files require SILDoulosIPA font
Length:28
Abstract:Positional Asymmetries and Licensing



Cheryl Zoll

MIT





The most restrictive version of Optimality Theory includes just three

families of constraints: Markedness, Faithfulness and Alignment (Prince

1997). Using only these constraints there have been two basic approaches

to licensing in phonology. In the first, licensing effects follow from

context-specific markedness constraints such as those in (1).



(1) Licensing is context-specific markedness: (Kiparsky 1995, Smolensky

1995, Ito and Mester 1996, Zoll 1996, Steriade 1997 inter alia)



License(voice): No voiced obstruents in coda position

License(complex segment): A complex segment coincides with the first

syllable



Alternatively, others have argued that licensing constraints are context

specific faithfulness constraints, exemplified by the POSITIONAL FAITH-

FULNESS constraint shown in (2).



(2) Licensing is context-specific faithfulness: (Selkirk 1994, Alderete

1995, Beckman 1995, 1997, Casali 1997)



IDENT-ONSET(voice): An onset segment is identical to its input

correspondent with respect to voicing



This paper, an expanded handout of a talk given at the LSA meeting in

January 1998, demonstrates that licensing must be due to context-

specific markedness constraints. The predictions of the two approaches

differ in at least three ways, and in each case only context-specific

markedness yields the correct result. In particular, the faithfulness

version of licensing: (i) wrongly restricts the possible repairs of

potentially unlicensed material; (ii) wrongly predicts that licensing

will not be sensitive to derived properties; and (iii) cannot straight-

forwardly account for necessarily negative licensing effects. It fol-

lows that while context-specific markedness constraints are indispens-

able, context-specific faithfulness is both inadequate and probably un-

necessary for licensing. This result has important consequences for

Optimality Theory more generally, since the expression of context-

specific markedness appears to require the additional power of con-

straint conjunction (Smolensky 1995).
Type:Paper/tech report
Area/Keywords:
Article:Version 1