|Title:||The Comprehension/Production Dilemma in Child Language: A Response to Smolensky|
|Authors:||Charles Reiss, Mark Hale|
|Abstract:|| The Comprehension/Production Dilemma in Child Language:
A Response to Smolensky
ROA-132 29 pp.
Charles Reiss and Mark Hale
Concordia University, Montreal
This paper constitutes in part a direct response to Smolensky (1996)
and in part a general contribution to the study of phonological acquisition.
Smolensky makes several claims in his paper, including those listed in (1):
(1) Major claims of Smolensky (1996)
(a) the "markedness" phenomena described in the child phonology literature
are paralleled by phenomena of adult phonology;
(b) speech output during direct imitation and the existence of chainshifts
in the mapping of adult language to child language argue for a competence-
(or grammar-) based approach to child speech production;
(c) an OT analysis can maintain that children possess a single grammar and
still capture the comprehension/production discrepancy in child language
(which is known to be difficult in a rule-based approach); and,
(d) Faithfulness Constraints (F-constraints) must initially be ranked
below Well-formedness constraints (W-constraints).
We will argue against points a-d, providing new arguments based on
evidence which is well-known in the linguistic, psycholinguistic and
phonetic literature. After sketching some basic assumptions about the
nature of phonological acquisition and its study, we will show that
the evidence actually supports the following claims:
(2) Our claims
(a) the supposed parallels between child and adult patterns of "markedness"
(b) improved performance under imitation is predicted only by a theory
which appeals to a performance basis; chainshifts are not a problem for
theories which take a "competence and performance" approach (and, in fact,
they remain unexplained in competence-only accounts, such as Smolensky's);
(c) the OT model of comprehension posited by S is empirically inadequate;
(d) within an OT framework learnability considerations favor an initial
ranking in which F-constraints outrank W-constraints.