Distinctiveness, Coercion and Sonority: A Unified Theory of Weight
 





Bruce Moren



University of Maryland, College Park











Cross-linguistically, phonological weight systems can be quite complex.
 

There are many factors to take into account when describing and 
 

explaining weight patterns, including:







1) segment type (consonant or vowel),



2) sonority,



3) weight distinctiveness,



4) prosodic weight requirements, and 



5) phonological environment (including syllable position).
 





The interaction of these factors can sometimes obscure relationships 
 

among them, thus leading to several separate approaches to account 
 

for different pieces of the weight puzzle.  For example, distinctive 
 

vowel length and distinctive consonant gemination are often seen as 
 

different phenomena to be accounted for in different ways.  Similarly, 
 

vowel lengthening and weight by position are seen as almost completely 
 

unrelated.  However, a thorough inspection of data from a large number 
 

of languages leads to the conclusion that a unified theory and 
 

mechanism of moraicity across segment types (i.e. both consonants and 
 

vowels) is warranted.  This work provides such a unified theory.
 





	The two main goals of this dissertation are:







1) to examine and review the nature and patterns of segment weight, 
 

including: inventories, processes, and dependencies; and
 

2) to provide a simple and economical account for the observed 
 

descriptive generalizations within the framework of Optimality Theory 
 

(OT) and Moraic Theory.







In addition, it becomes clear throughout this work that one major 
 

advantage of OT over traditional rule-based theories is the 
 

typological predictions intrinsic to the architecture.  Since OT 
 

constraints are universal, violable and re-rankable, their factorial 
 

ranking potentially yields all possible grammars.  In the case of 
 

segmental weight, the factorial ranking of three simple constraint 
 

families yields the major descriptive generalizations found 
 

cross-linguistically.







	Chapter 1 reviews evidence for different degrees of weight, 
 

presents the syllable representations assumed throughout this work, 
 

and demonstrates that there are two sources of weight - coerced and 
 

distinctive.  Coerced weight is a restriction on surface moraicity in 
 

some phonological context (e.g. weight by position and foot binarity), 
 

and is subject to distributional restrictions based on sonority. In 
 

contrast, distinctive weight is an underlying moraicity reflected in 
 

a surface contrast (e.g. geminate versus non-geminate intervocalic 
 

consonants), and is not bounded by sonority.  Both sources of weight 
 

are observed to be relevant to both vowels and consonants.
 





        Chapter 2 briefly reviews Optimality Theory and Correspondence 
 

Theory, discusses the relationship between OT and typology, and 
 

presents the factorial rankings (permutations) of three types of 
 

constraints: 







1) General moraic markedness constraints against moraic segments of 
 

different types - ranked in a universal hierarchy based on sonority;
 

2) Coercive moraic markedness constraints; and



3) Faithfulness constraints on underlying moraic affiliation with 
 

segments of different sonorities. 







        Chapter 3 uses data from a number of languages to show that 
 

the descriptive generalizations discussed in chapter 1 emerge 
 

naturally as the result of interactions among the constraints 
 

formalized in chapter 2.







        Chapter 4 expands on chapter 3, and provides in-depth case 
 

studies of segment moraicity and other phenomena in Hawaiian, Modern 
 

Standard Italian, Kashmiri, two Hungarian dialects, two Icelandic 
 

dialects, and Metropolitan New York English.  This chapter gives 
 

detailed descriptions of different weight patterns; reveals that the 
 

constraints proposed in this work can be integrated into more complete 
 

grammars, and shows that different dialects can arise from a minimal 
 

re-ranking of constraints.







        Chapter 5 is a repository for miscellaneous issues, as well as 
 

the general conclusions.  Specifically, it provides a discussion of 
 

why there is no need for constraints that target long vowels and/or 
 

geminate consonants to derive inventories; it addresses the work of 
 

Broselow, et al (1998) regarding phonetic reflections of phonological 
 

weight; it discusses the advantages of using negative moraic 
 

markedness constraints over positive constraints; and it shows that 
 

Tranel's (1991) Principle of Equal Weight for Codas is the logical 
 

result of constraint interaction.