|Title:||Consonant Harmony in Child Language: An Optimality-theoretic Account|
|Abstract:||Consonant Harmony in Child Language:
Department of Linguistics, McGill University
Ms. dated 1995. Currently in press in S.J. Hannahs & M.
Young-Scholten, _Focus on phonological acquisition_. Amsterdam:
John Benjamins, pp. 113-142.
An analysis is provided of the consonant harmony (CH) patterns
exhibited in the speech of one child, Amahl at Stage 1 (data from
Smith 1973). It is argued that the standard rule-based analysis
which involves Coronal underspecification and planar segregation
is not tenable. First, the data reveal an underspecification
paradox: coronal consonants are targets for CH and should thus be
unspecified for Coronal. However, they also trigger harmony,
in words where the targets are liquids. Second, the data are not
consistent with planar segregation, as one harmony pattern is
productive beyond the point when Amahl's grammar satisfies the
requirements for planar segregation (set forth in McCarthy 1989).
An alternative analysis is proposed within the constraint-based
framework of Optimality Theory (Prince & Smolensky 1993). It is
argued that CH follows from the relative ranking of constraints
which parse place features and those which align features with
the edges of the prosodic word. With the constraints responsible
for parsing and aligning Labial and Dorsal ranked above those
responsible for parsing and aligning Coronal, the dual behaviour
of coronals can be captured. The effect of planar segregation
follows from other independently motivated constraints which
force alignment to be satisfied through copying of segmental
material, not through spreading.
Keywords: first language acquisition, consonant harmony,