LINGUIST List 5.811

Thu 14 Jul 1994

Review: The Development of Speech Perception

Editor for this issue: <>


Directory

  1. Helen Dry, review

Message 1: review

Date: Wed, 13 Jul 1994 10:34:06 review
From: Helen Dry <hdryemunix.emich.edu>
Subject: review


Discussion/Review: The Development of Speech Perception:
The Transition from Speech Sounds to Spoken Words, edited by
Judith C. Goodman and Howard C. Nusbaum, The MIT Press,
1994. $40.00

 The field of research covered by The Development of
Speech Perception (henceforth DSP) has only recently begun
to receive the attention it deserves. Although child
language has played a role in linguistic theory since
Jakobsen's "Kindersprache, Aphasie, und allgemeine
Lautgesetze", there has been a strong tendency to consider
the speech of children as something structurally distinct
from that of adults, almost as if children spoke a
different language. Research has tended to concentrate
either on infant or adult speech perception, and has
generally neglected to examine the process of change from
one "stage" to the next. Taking a developmental perspective
on the process of speech perception not only allows the
authors of the chapters in DSP to examine questions
pertinent to speech perception research, but also provides
challenges to some common assumptions of the currently
standard paradigm of linguistic research.
 The book contains nine chapters, the first of which
("Developing Theories of Speech Perception: Constraints on
Developmental Data", by Judith C. Goodman, Lisa Lee & Jenny
DeGroot) provides an introduction to and overview of the
book, and the last of which ("Learning to Hear Speech
as Spoken Language", by Howard C. Nusbaum & Judith C.
Goodman) serves as a sort of afterword and suggestion for
further research, although I suspect that not all of the
authors of the other chapters would agree unconditionally
with the conclusions drawn.
 Chapters two through nine are organized into three
parts. Chapters two ("Observations on Speech Perception,
Its Development, and the Search for a Mechanism", by Joanne
L. Miller & Peter D. Eimas) and three ("The
Importance of Childhood to Language Acquisition: Evidence
from American Sign Language", by Rachel I. Mayberry) make up
Part I, "Innate Sensory Mechanisms and Constraints on
Learning". Part II, "Perceptual Learning of Phonological
Systems" consist of three chapters, "Cross-Language Speech
Perception: Developmental Change Does not Involve Loss" by
Janet F. Werker, "Perceptual Learning of Nonnative Speech
Contrasts: Implications for Theories of Speech Perception"
by David B. Pisoni, Scott E. Lively & John S. Logan, and
"The Emergence of Native-Language Phonological Influences in
Infants: A Perceptual Assimilation Model" by Catherine T.
Best. Part Three, "Interaction of Linguistic Levels:
Influences on Perceptual Development", contains the chapters
"Infant Speech Perception and the Development of the Mental
Lexicon" by Peter W. Jusczyk, and "Sentential
Processes in Early Child Language: Evidence from the
Perception and Production of Function Morphemes" by LouAnne
Gerken, as well as the above mentioned chapter by Nusbaum
and Goodman.
 This book is the outcome of a Workshop on Recognizing
Spoken Language he,dw the University of Chicago in June
1989. As noted in the Preface "The goal of this workshop
was to examine transitions in the perceptual processing of
speech from infancy to adulthood...In particular the
participants were invited to speculate about how their
findings constrain the nature of the mechanisms and
representations that mediate speech perception during
infancy, childhood, and adulthood." Thus, though each of
the chapters provides a summary of data from years of
research in infant and/or adult speech perception by the
author(s) and their colleagues, the primary focus of the
book is theoretical. The authors come from diverse
backgrounds and theoretical perspectives, and it is clear
from the outset that the theoretical conclusions drawn by
each author or group of authors will not necessarily agree,
however there seems to have been a great deal of cross-
fertilization, and many chapters make at least a significant
passing reference to work by other authors included in the
volume.
 As the title of chapter nine, "Learning to Hear Speech
as Spoken Language", suggests, the research presented in
this book demonstrates that the process of speech perception
is considerably more than a simple decoding of the speech
signal serving as a front-end to language understanding.
Rather, Nusbaum & Goodman present an excellent set of
arguments against the prevailing modular theory of language,
in which the "performance-based" fields of speech perception
and acoustic phonetics are contrasted with the "competence
oriented", linguistic fields of phonology, morphology,
syntax, and semantics. This distinction has been maintained
by the fiction that speech perception is merely a bottom-up
transduction system whose purpose is to "render the messy,
highly variable acoustic waveform of speech into the
pristine, discrete, symbolic form that can be dealt with by
higher-order psycholinguistic processes." (p. 302). This
division is contradicted by recent research "suggesting that
the mechanisms of speech perception are coextensive with, in
i]service of, and dependent upon the mechanisms of language
understanding in fairly complex ways." (p. 303).
 The field of cognitive psychology provides a framework
for Nusbaum & Goodman to explore speech perception as an
integral part of language understanding, and as a process
that develops in a continuous manner in concert with the
development of all other cognitive abilities. Citing
Nusbaum & Henly (in press), Nusbaum & Goodman argue that
"speech perception is not a simple recognition process that
maps acoustic properties onto linguistic categories, and
development is not a simple process of determining which
mappings are appropriate for a particular language." (p.
320). They suggest that "[i]nstead, speech perception and
spoken language understanding is a process of constraint
satisfaction and hypothesis testing in which the particular
constraints that are available and the hypotheses being
tested vary dynamically with aspect of context and the
communicative situation." (320). Nusbaum & Goodman rely
heavily on the concept of directing attention to target
linguistically useful aspects of the speech signal. The
direction of attention is based on all sorts of knowledge
about the signal, including context, knowledge of the
language, and knowledge of the speaker. Thus linguistic
categories (such as phonemes or words) cannot be represented
mentally in terms of fixed lists of features or properties,
because such a list would have to account for all of the
possible variations in characteristics or features depending
on all possible contexts. Such a list of features could
exist, though it would be ridiculously long, if it weren't
for the fact that some information used in speech
perception, such as the role of lexical knowledge in phoneme
perception, or that of the particular relationship between
the vowels of each individual talker's vowel space in
identifying vowels, cannot be easily incorporated into a
list of features. Instead of feature lists, Nusbaum &
Goodman propose that linguistic categories be though of as
"theories that people have about the structure, nature, or
function of those categories." (p. 322). Recognition is
then a process of testing these theories against the data
available in the signal. This is an interactive process
that is inherently flexible and adaptive. Such a
cognitive theory of speech perception allows one to account
for both the development of speech perception as a
continuous modification, from infancy to adulthood and
including learning new languages, of the theories of
categories, and successfully integrates the process of
speech perception, which is obviously not only a
"performance" phenomenon, into the realm of "cs3?vtence"
based language understanding.

 afrancismidway.uchicago.edu alex francis
(312)-539-4451

 "Freiheit ist immer die Freiheit des
Andersdenkenden."
 - Rosa Luxumbourg
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue