|Title:||Reinterpreting Transparency in Nasal Harmony|
|Abstract:||Reinterpreting Transparency in Nasal Harmony
University of Southern California
This paper presents a unified account of nasal harmony (NH). Three
segment behaviors are examined: 'targets' become nasalized in NH,
'blockers' remain oral and block spreading, and 'transparent segments'
remain oral but do not block. Blocking patterns have previously been
observed to obey a bifurcational hierarchy, separating possible targets
(leftward) versus blockers (rightward) (e.g. Schourup 1972; Piggott
1992; Cohn 1993; also Pulleyblank 1989):
(1) i. Vowels ii. Glides iii. Liquids
iv. Fricatives v. Obstruent Stops (vi.)
The present study extends this nasalization hierarchy to all NH. A
complementarity provides motivation: there appear to be no examples for
(vi), where all segments, including obstruents are nasalized, and another
system is set apart from the others, one in which some obstruents act
transparent and remaining segments are targets. Also, target and
transparent segments are discovered to pattern together with respect to
the hierarchy in (1): if nasalization 'permeates' a segment (it is
targeted/transparent), all higher-ranked segments are permeated. To
explain this, descriptively transparent segments are analyzed as targets
of NH, following proposals that feature spreading occurs only between
strictly adjacent segments (Gafos 1996; Ni Chiosain & Padgett 1998).
The nasalization hierarchy is then reinterpreted as representing possible
bifurcations between 'permeable segments' and blockers. Systems with
transparency--where all segments are permeable--are slotted into step
(vi), achieving a unified typology of NH where the hierarchy exhaustively
limits variation. The scaling in (1) is formalized in Optimality Theory
as a fixed nasalized segment constraint hierarchy. Ranking a [Nasal]
spreading constraint at each point in relation to this nasalization
hierarchy gives the crosslinguistic variation, yielding typology from
factorial ranking (Prince & Smolensky 1993).
Permeable segments have two possible phonetic outcomes: nasal or oral.
The latter occurs on permeated segments near the extreme of
incompatibility with nasalization (e.g. voiceless obstruent stops).
This 'antagonistic transparency' (Archangeli & Pulleyblank 1994) is
analyzed as a 'derivational opacity' effect (in the sense of Kiparsky
1973). Generative antecedents (e.g. Vago 1976, Clements 1976) posit full
spreading at an abstract level of representation followed by application
of a rule deleting the feature on surface-transparent segments. The OT
analysis proposed here follows these accounts in calling on a fully-
spread representation; however, it employs only one level of input-output
mapping by drawing on co-candidate correspondence, known as 'Sympathy'
(McCarthy 1997; with extensions by Ito & Mester 1997). The derivational
effect arises in a correspondence mapping (Ident[Nasal]; McCarthy &
Prince 1995) between a fully-nasalized but unpronounceable phonological
output representation and a similar but phonetically-possible output.
This account thus achieves the phonetic discontinuity produced by
antagonistic feature combinations, while maintaining strict segmental
locality for feature linkage--segmental transparency is brought under the
umbrella of widespread derivational opacity and transparency-specific
representations, such as gapping, are obviated in the theory.