LINGUIST List 15.683

Tue Feb 24 2004

Review: Lang Acquisition,Sign Lang:Goldin-Meadow (2003)

Editor for this issue: Naomi Ogasawara <naomilinguistlist.org>


What follows is a review or discussion note contributed to our Book Discussion Forum. We expect discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for review." Then contact Sheila Dooley Collberg at collberglinguistlist.org.

Directory

  1. Annette Hohenberger, Resilience of Language

Message 1: Resilience of Language

Date: Tue, 24 Feb 2004 14:06:01 +0100
From: Annette Hohenberger <hohenbergercbs.mpg.de>
Subject: Resilience of Language

AUTHOR: Goldin-Meadow, Susan
TITLE: The Resilience of Language
SUBTITLE: What Gesture Creation in Deaf Children Can Tell Us About How
 All Children Learn Language
PUBLISHER: Psychology Press (Taylor and Francis)
YEAR: 2003

Announced at http://linguistlist.org/issues/14/14-3118.html


Annette Hohenberger, Max-Planck-Institute for Human Cognitive and
Brain Sciences, Department of Psychology, Munich

INTRODUCTION

In this monograph, Susan Goldin-Meadow explores the "resilient"
properties of language, i.e., those properties which are robust and
flexible enough to occur universally (across the languages of the
world) in individuals under vastly varying acquisitional
scenarios. The "language-making skills" (rather than language-learning
skills) of young children can access grammatical forms and
constructions implementing these properties easily in normal as well
as extremely unfavorable learning situations.

In order to identify the resilient properties of language, Susan
Goldin-Meadow studies deaf children of hearing parents who have
neither a sign nor a spoken language model from which they could
derive the properties of human language. Despite this lack of a
language model, these subjects come up with a language system that has
all the hallmarks of natural language.

What exactly are resilient properties? Susan Goldin-Meadow calls those
properties of language resilient whose development is resistant to
changes in the environmental conditions under which languages are
learned: "To the extent that a property of language is UNAFFECTED (in
italics in the original, my remark, A.H.) by a given manipulation, it
can be said to be developmentally RESILIENT - its developmental course
is impervious to the change in input conditions." (p 19) The opposite
of resilient properties are "fragile" properties "whose development is
sensitive to changes in input conditions." (p 20)

The question of what language children would develop if they were
deprived of language has its historical antecedents in the search for
the proto-language of mankind. It is historically bequeathed that
several emperors, namely the Pharaoh Psammetic, the Staufer King
Frederic II and the Scottish King Jacob IV arranged for rigorous
experiments on newborns which they had deprived of any language and
human companion in order to find out what language they would
develop. This language should then be considered the human
proto-language. These experiments all failed: the poor infants either
died (as in Frederic II's case) or uttered only some sparse
proto-words (which, however, led Psammetic to conclude that Phrygian
must be the proto-language and Jacob that it was Hebrew). Thus, the
most basic condition which must be met for any language-learning to be
possible at all is human companion and the willingness to communicate,
i.e, "human and humane contact." (p 48) This is the case for the
subjects Susan Goldin-Meadow studies. They live under normal social
and emotional conditions except that they do not have a language model
because (1) their hearing parents cannot sign and (2) they want them
to acquire an oral language, spoken English. As a result neither the
parents get what they want - a speaking child - nor do the children
get what they need - a(ny) language model, signed or spoken. The study
therefore qualifies as a "deprivation study" conducted under the only
ethically acceptable condition - that it be not deliberately created
by the experimenters. Rather, the author came across these children
in their specific situation which she identified as offering "a unique
and powerful way to isolate properties of language that are
resilient..." (p 51)

SUMMARY

The book is organized into three main sections:

In the first section (chapters 1-5) the author lays out "The problem
of language-learning". In this section, she covers the common
developmental sequence children go through despite cross-linguistic
differences, despite differing circumstances of acquisition, and
despite different modalities (aural-oral for spoken languages;
visual-gestural for sign languages).

In the second section (chapters 6-15) she explores in depth the
situation of "Language development without a language model". This
section is the core section which covers a wealth of empirical data
and very thoroughly carried-out studies on the structure and use of
gestures which the deaf children in her study create in the absence of
any model of language. Furthermore, she compares the gestures of the
hearing parents with those of their deaf children and also the
gestures of deaf children from another culture/language, namely
children from Taiwan whose surrounding spoken language is
Chinese. Those comparisons should factor out any influence of the
parents' gestures and of the culture and show which properties are
really resilient and which can be modified by gestural input and
cultural impact.

In the four chapters (16-19) of the final section "The conditions that
foster language and language-learning" she generalizes her findings
with respect to the question of how language-learning proceeds in
children in general and what role the resilient properties of language
and the language-making abilities of children play. She also studies
gesture-creation experimentally in hearing adults. From all these
studies she derives a new perspective on the longstanding and
hotly-debated question whether language is innate by shifting the
emphasis from genetics to developmental resilience.

This book, according to the author, aims at students of linguistics at
the upper level undergraduate courses and introductory graduate
courses. It can be used for teaching, though not as a stand-alone
textbook but in combination with more traditional textbooks on
language acquisition. The book can be read (and taught) in a modular
way. The sections and individual chapters can be read independently of
each other. Especially the first section is a concise introduction
into the fundamental problems of language acquisition research and
discusses the various theoretical approaches on the background of a
wealth of empirical findings. Readers already acquainted with the main
paradigms, problems, and puzzles of language acquisition may rehearse
their knowledge by reading this section before concentrating on the
central second section. In each of the chapters in the second section,
particular aspects of the deaf children's gesture system are
discussed. In class, they can be supplemented by the more detailed
original studies they are taken from and which have already been
published as independent papers. In the last section, conclusions are
drawn from the empirical studies, conventional language acquisition is
revisited in the light of the present findings, speculations on the
evolution of gestural and oral languages are sketched, and the
relation of gesture and speech in language is discussed.

The book is supplemented by 17 movies which show in real-time the
gestured words and sentences of the deaf children which are referred
to as examples in the book. The movies are available at
http://www.psypress.com/goldinmeadow

There are a few typos which are hardly worth mentioning
('manipulatable' on p 63 and 'old' instead of 'odd' on p 70). What
should be corrected in the next edition is a mislabelling in the
legends of two figures. In Figure 17 (p 139) the 'distal events'
should be in white and the 'object & action knowledge' in black. In
Figure 23 (p 174) the 'American children' should correspond to the
black bars and the 'Chinese children' to the striped bars. Everything
is fine in the diagrams themselves, only the legends are incorrect.

At the heart of this monograph is the description of the gesture
system of 10 US deaf children of hearing parents from 1;4 (earliest
session) to 4;6 (latest session) years of age. All utterances were
scored according to a coding-system which was developed for the deaf
children's utterances and which was also used for the other groups
with which the original subject group was compared later. The
coding-system consists of three categories, namely "deictic (pointing)
gestures" which function like nouns (N) and pronouns and denote
persons/objects/locations, "iconic gestures" which denote actions and
events (either transitive or intransitive), and "marker/modulator
gestures" (e.g., headnods, headshakes) which convey mode
(affirmation/negation/doubt/want) (chapter 6). The meaning of these
gestures was determined on the basis of their form. Inter-rater
reliability was sufficiently high (87-100%). Gestures were elicited
from the young subjects by an experimenter in interactive play
sessions.

The gestures the children created and used are clearly organized as
parts of a language system. This language system displays all signs of
a natural language. It includes a lexicon in which words - nouns and
verbs - are stored as stable lexical entries relating form and meaning
in a systematic way, a morphological component which creates
paradigmatic contrasts of e.g., handshapes and movements, predicate
frames specifying "who does what to whom", and, finally, a syntactic
component with recursive capacities, serving all the functions natural
language is used for: making requests and comments in the
here-and-now, referring to non-present objects and events, making
generic statements, telling stories, talking to one-self and, last,
serving meta-linguistic purposes.

Some forms and meanings seem to be more easily accessible in language
acquisition than others. The most compelling of those are "ergative"
constructions, a recurring theme throughout the book (see also
Goldin-Meadow 2003a). The deaf children focus primarily on patients
and on the result of actions as they are evidenced by the equally high
production probabilities of intransitive actors and transitive
patients in the deaf children's data as opposed to the low production
probability of subjects of transitive verbs. Underlying this pattern
is "a tendency to see objects as affected by action rather than as
effectors of action." (p 106) As in ergative languages, where
intransitive actors and transitive patients are marked alike
morphologically, the deaf children group them together by means of the
same probability of occurrence. Furthermore, they are also treated on
a par with respect to word order: both transitive patients and
intransitive actors precede the acts, as in "snack EAT" and "Susan
MOVE-OVER". Transitive actors, if they appear at all, occur
post-verbally. From these results one can deduce that the
spontaneously emerging word order for the majority of the deaf
children is OV(S) and that they have acquired an ergative
language. Not only the deaf American children show this pattern, but
also deaf Chinese Children who grow up under comparable conditions of
language deprivation (chapter 15). Hearing adult subjects, too, come
up with ergative (and figure-ground) word order if asked to describe
scenes by means of gestures only (chapter 17). The pervasive ergative
pattern may result from more general cognitive factors in parsing
events into units of speech or sign (see below). Languages can and do
override this initial option which is part of a "gold standard" (p
173) in acquisition and processing. Multiple factors, cognitive,
linguistic, communicative, interact in explaining word order and event
packaging in a given language. Lacking any language model, the deaf
children show this gold standard in its pure form. Carrying the gold
metaphor a little further, they can be said to have spun straw into
gold - the proven structure- and worthless gestural input of their
parents (chapter 14) into a natural language.

Susan Goldin-Meadow summarizes the resilient properties of language
she found in her deaf child subjects in chapter 16, Table 6 (p 186).
Resilient properties of words are their stability in form, their
segmentation and combination into morphological paradigms consisting
of a limited set of categories (handshapes and movements), the
arbitrariness of the form-meaning pairing (despite a certain degree of
iconicity) and their differential functioning as nouns, verbs, and
adjectives. The resilient properties of sentences are underlying
predicate frames with consistent production probabilities of
particular theta-roles, word order (OV), inflections, recursion, and
redundancy reduction. In Table 6, words and sentences are taken as
domains in which resilient properties of language manifest
themselves. Making this point even stronger, one might consider
morphemes, words, and phrases themselves as resilient units of
language which emerge in a predictable way through processing and
acquisition. The resilient functions of language comprise here-and-now
talk, displaced talk, generics, narrative, self-talk, and
meta-language. In the same vein, she identifies language-making
skills that do not require a language model, as summarized in Table 7
(p 190). Processes which are available despite a missing language
model are segmenting (words into morphemes, sentences into words),
constructing (morphological) paradigms and constructing syntactic
sequences. Thus, segmentation and combination (on the vertical and the
horizontal axis) are so basic that they can occur de novo in the
course of language acquisition. Structures which can be attained
without a language model comprise multiple argument predicate frames,
the grammatical categories of nouns, verbs, and adjectives, and
ergative constructions. The overall acquisition path of the deaf
children is also comparable to those of children with a language
model, namely from the whole to the parts (chapter 9). They start out
with unanalyzed word forms which represent primarily the gesture-world
relation. Once they have stored a critical mass of entries in their
mental lexicon, they start breaking down these forms, extract
morphemes and recombine them, thereby creating paradigms of systematic
contrasts which are now governed by gesture-gesture relations, i.e.,
by language-systematic aspects. As for syntax, they start out with
single gestures, then combine them into two-gesture sentences
(1;6-2;5) and, finally, at the age of (3;1-3;11), they produce complex
sentences. In this process of "building a system" (chapter 12) they
re-analyze or "redescribe" their representations, to speak with
Karmiloff-Smith (1992). Evidence for that also comes from their
shifting from semantic notions such as agents and actors to notions of
grammatical functions such as nouns and verbs. One deaf child's
development was monitored in particular in this respect. He started
out with a semantically-based correlation of objects with nouns and
actions with verbs. Before the age of 3;3 nouns and verbs were marked
strictly formally in that nouns were always deictic gestures and verbs
iconic gestures. After 3;3, this equivalence relation was resolved
stepwise. As a first interim step, he then expressed nouns with iconic
gestures, too, but never used the same gesture for both nouns and
verbs. This amounts to a lexical strategy. Finally, nouns and verbs
became truly grammaticized notions. He then used the same iconic
gestures for nouns and verbs but distinguished them by morphological
markings. Susan Goldin-Meadow argues that there is a developmental
nexus of grammaticalization of semantic notions, the emergence of
morphology (derivational and inflectional) and, maybe, also the
symbolic use of gestures for non-present objects (chapter 12). Apart
from these compelling parallels in the developmental trajectories of
children with and without language models, there are also
differences. Not surprisingly, the development of the latter is
somewhat delayed and their sentences remain shorter, in general.

EVALUATION

"The resilience of language" is a book which takes us back to the very
roots of language in terms of acquisition and evolution. It attempts
at answering a question which is as old as our reasoning about our own
nature: What is at the very heart of our human language faculty?

Susan Goldin-Meadow's methodological approach is sound and sharp and
therefore allows her to establish reliable and valid evidence with
respect to her research question. She adheres to the following
rationale: "We first establish an equivalence between the deaf
children's systems and the linguistic systems developed by children
learning conventional languages. We then make inferences about how
language is learned in general from the fact that the deaf children
display certain properties of language in their systems but fail to
display others." (p 123) She proceeds very carefully in trying to
exclude possible intervening hypotheses. By showing that the gestures
of the deaf children's speaking parents have no language-systematic
properties at all, she excludes the reasonable assumption that the
parents' gesturing may have fostered their children's gesturing
(chapter 14). By showing that another group of deaf children who grow
up in Taiwan (whose parents speak Chinese) show the same syntactic
patterns than do their US counterparts, she proves that neither
culture nor the surrounding language has any influence on the
structural patterns which emerge in deaf children without a language
model (chapter 15). By showing that English speaking and Chinese
speaking peers do show salient differences in overt realization of
arguments of underlying predicate frames, she identifies those aspects
of language which can only develop given a particular language model
(chapter 15). These properties are the "fragile" properties of
language. By showing that adult speakers, if asked to rely on gestures
only in describing dynamical scenes, resort to the same word orders
and event packaging strategies as the deaf children did, she
identifies a common cognitive substrate for these recurring phenomena
(chapter 17).

The properties summarized in table 6 and 7 are called "resilient
properties of language" and resilient language-making skills,
respectively. The question is: Are they really resilient properties of
language or are they resilient properties of cognition? If the former
is true, then these properties are specific to language and do not
show up in a different cognitive domain. If the latter is true, then
these properties hold for cognitive development in general. Thus, the
question is if what Susan Goldin-Meadow has identified
phenomenologically is resilient with respect to language or resilient
with respect to cognition in the domain of language or, even wider,
communication. Are there general (developmental) cognitive schemas and
strategies which, if applied to the task of communication, produce the
phenomena at hand?

In order to examine this possibility, let us exemplarily come back to
the prominent ergative pattern. I would like to adduce evidence from
findings in cognitive and developmental psychology which are strongly
reminiscent of this pattern. In cognitive psychology, it is well-known
that the (anticipated) results of actions play a key role in the
perception, planning, and control of actions. This focusing on the
results or goals of actions has been dubbed the "action-effect"
principle (Prinz 1997). In developmental psychology, too, it has been
shown that infants pay special attention to salient action effects on
the objects involved in human action (Jovanovic et al. submitted). In
both domains, human action and language, there seems to be a common
bias for the encoding of results/end states/patient properties of
events.

Likewise, for the other resilient properties and processes, one
wonders if they are cognitive in general or language-specific. Susan
Goldin-Meadow discusses this central question (chapter 16, p 191f)
when she asks why the deaf children co-opt for specific processes,
namely segmentation and combination of units into paradigms and
syntactic sequences, thereby following underlying predicate frames,
and introducing recursion. In particular, she asks why the deaf
children did not grammaticize e.g. Euclidean-geometric concepts such
as distance, size, contour, angles, etc. which would be natural given
the spatial nature of their gesture system. Instead, they grammaticize
the very same notions as do all natural languages. She concludes:
"That the children choose to co-opt the particular combinatorial
processes listed in Table 7 over others is therefore significant. Even
if we are completely comfortable saying that these processes reflect
general cognitive operations (as opposed to specifically linguistic
operations), we still need to explain why THIS SET (in italics in the
original, my remark, A.H.) has been co-opted." (p 192) In her answer
she invokes the task requirements of language itself: "It may be that
the communicative situation itself demands not only segmentation, but
also combinatorial processes of this sort - ..." (ibid.) The task of
transmitting information requires that our "knowledge must go through
a discrete filter and, as a result, its representation ends up looking
discrete." (p 191) However, this answer is only partially satisfying
as there is no (and can be no) causal explanation of how these task
requirements shall result in grammatical structures. They can only
define the problem and set the goal, here: how to get one's thoughts
into the head of someone else and vice versa. There must be processes
and structures ready for implementing this task. It is these processes
and structures the child co-opts. Thus, we are back where we started.

If what the children developed on their own is a result of their
general cognitive abilities, then what is specific to language as a
module (if one does not want to deny its status as a cognitive module
altogether)? In order to answer this question, one might look at what
properties the deaf children did not develop but which are considered
crucial for language, though. What comes to mind first, is
phonology. Susan Goldin-Meadow says nothing about the analysis of
gestures into the three (or four) phonological parameters of
conventional sign languages, namely handshape, movement, place of
articulation (and hand-orientation). Her deaf children use the first
two of them, handshape and movement, for decomposing words into their
constituent morphological components. Implicitly, phonological
features must be accessible to them, but maybe only via
morphology. She says nothing about systematic phonological contrasts
of the other two features. Either there is too little evidence for
them in the data or the gestures really have not been analyzed
phonologically. Given that phonology is a defining characteristics of
all natural languages, it is hardly conceivable that they haven't. A
phonological analysis might even require more time to develop in
acquisition (and evolution) than a morphological analysis, given the
whole-part strategy mentioned above. The hearing adults in the
"gesture creation paradigm" also neither developed a morphology nor a
phonology.

In the domain of syntax, too, the deaf children did not come up with a
full-blown syntactic system. Susan Goldin-Meadow grants them the very
basic property of recursion. Thus, they combine phrases to complex
sentences, although they cannot show their full recursive capability
of multiple embedding, given the overall shortness of their
sentences. The lack of complementizers and hypotactic subordination is
probably also due to this basic limitation. There is, however, one
property missing which cannot be explained away along these lines. Of
the two basic operations of syntax they have Merger but not Move. Move
relates to the basic property of displacement and formation of chains
which is common to all natural languages (at least from a generative
perspective). The three basic layers of a phrase marker, the thematic
layer (VP), the propositional layer (IP), and the modal layer (CP) are
related to each other by movement of constituents and creation of
chains. There is no evidence in the data of the deaf children that
they, for example, topicalize a constituent by fronting it. It is hard
to conceive how they should figure out such a possibility, lacking any
language model. However, dislocation is an 'essential' property of
language.� Maybe it is even a resilient one, but one which might
need more time to develop, and a wider range of application.

It might be helpful to not only contrast 'resilient' with 'fragile'
properties but also with 'essential' and 'emergent' properties of
language. 'Essential' properties of language would be those which are
specific to and constitutive of language, e.g., features like TNS or
processes like Move which have no counterpart in any other cognitive
system. Essential properties of language may or may not be
resilient. I would suggest to distinguish those resilient properties
which are domain-unspecific from those which are domain-specific. If
it turned out that there exist only domain-unspecific resilient�
properties of language, this would be tantamount to the connectionist
claim that there are no language-specific properties at all. If there
were language-specific resilient properties, however, this would be
strong evidence for those aspects of UG which are data-insensitive. It
is hard to clear the notion of 'resilience' from being 'essential',
'most basic', or 'most important' in language. It may turn out that
essential properties of language are never resilient and manifest
themselves only in connection with input whereas more general
cognitive strategies are responsible for the resilient
properties. Note that this scenario would not be incompatible with a
theory of UG which holds that UG only defines the boundary conditions,
i.e., the possible format of human language and not the set of
linguistic primitives per se, namely "if units of representation are
allowed to emerge from the accommodation of actual data according to
general principles of the representational system." (Bierwisch 2001:
299)

Structures may also arise in a process of 'emergence' rather than
learning from a model. 'Emergent' properties are novel systemic
properties which arise spontaneously ('de novo') on a macroscopic
level in the course of (language) development, due to local
interactions of elements on a microscopic level. Morphemes, words and
phrases as recurrent and stable units of language, N and V as
grammatical categories, and word order, are good candidates for
emergent structures. Resilient structures might therefore be emergent
ones.

Truly fragile properties are branching direction (left or right, see
chapter 11) as an example of a classical parameter in the sense of the
Principles-and-Parameters Theory. They can be easily fixed by limited
experience but a minimal amount of experience must be available. A
single person cannot be the creator of language and provider of
evidence for him-/herself at the same time, as correctly pointed out
by Susan Goldin-Meadow. This would amount to an absurd bootstrapping
task. In order for language to fully bloom other conditions must be
met, as in the felicitous case of the emergence of Nicaraguan Sign
Language (NSL) which she discusses in chapter 19. First, there must be
a sufficiently large number of members of the language community who
are willing to communicate, second, there must be the opportunity to
see the system as a whole which is true for new members who enter this
community, and, as a third requirement, they must be of a sufficiently
young age. Of all these preconditions, only the last is true for the
deaf children of the present study. They did the best they could and
got as far as possible in their given situation.

Susan Goldin-Meadow also discusses other accounts of language creation
in similar situations in order to clarify her own account of
developmental resilience (chapter 5). Similar evidence for the
language-making skills of children come from Bickerton's (1999) creole
studies. He invokes a "bioprogram" operative in young children which
enables them to enrich the pidgin they receive as input and to develop
it into a structurally complex creole language. One might be tempted
to simply subsume the resilient properties of language under the
innate language bioprogram. There are two main differences between
the two accounts, though. First, Susan Goldin-Meadow's methodology and
data are even more radical and therefore also more evidential for the
identification of the resilient properties of language than
Bickerton's data. Second, whereas Bickerton readily resorts to
nativism as an explanation, Susan Goldin-Meadow is much more cautious
in this respect. She carefully avoids any direct claims of innateness
or biological anchoring of the human language faculty. This abstinence
makes her approach especially valuable as a concept that can be shared
by different learning approaches - innatist, connectionist,
environmentalist - which she outlines in chapter 2. Rather than
defining "innate" as genetic she suggests "that the definition is more
usefully anchored in the notion of developmental resilience."
(chapter 18, p 215) With developmental resilience, she means
developmentally "buffered against a large number of both environmental
and organic variations." (p 218) Language acquisition can succeed
under a vast range of differing conditions, even under conditions of
no input at all. Some aspects of language are more affected by these
environmental variations than others. The latter are "central to
humans - so central that their development is virtually guaranteed,
not necessarily by a particular gene but by a variety of combinations
of genetic and environmental factors. In this sense, language is
innate." (p 220) Her account of innateness in terms of developmental
resilience is suited to relax the ongoing innateness controversy
(Elman et al. 1996) in that it avoids the paradigmatic dispute between
the various theoretical camps but focuses more on the content of what
is claimed to be innate. She has provided a detailed catalogue of
resilient properties of language based on a thoroughly cared-out
empirical study of a group of language learners under the most hostile
conditions one can think of. Even without any model of language, the
developmental path is not random but runs along the same general
trajectories as in any other group of language learners. Language
development is "equifinal" (p 218) although it obviously is not
equi-initial or equi-liminal (Hohenberger 2002), i.e., it can start
out from very different initial states and take different possible
pathways from the initial to the steady state. Susan Goldin-Meadow has
shown how substantial variation can be in the initial conditions and
still all these trajectories lead to a single stable solution within
an overall dynamical language system - due to the resilient properties
of language. Some of these properties may take more or less time to
become manifest. Some pop up de novo in a single generation, others
may take considerably more time to develop and transcend the
ontogenetic time-span. In chapter 17 she speculates on the role of
time in the manifestation of resilient properties. Time, however, is
also confounded with linguistic evidence. The next generation would
be provided with the evidence built up by the previous
generation. Linguistic features emerge through the iterated process of
mapping external data on internal representations (grammar). The
longer this process operates, the subtler and more "fragile" these
features will be. All features, resilient and fragile ones, are
licensed by the same conditions of the human language module but have
differing developmental onsets. The author has explored the very
beginning of this process, the first few iterations of a dynamical
process in an ontogenetic time-span. Language develops on various
time-scales at the same time - evolutionary, phylogenetic,
ontogenetic, microgenetic. Resilience manifests itself on all these
scales but not necessarily for the same properties at the same speed.

Susan Goldin-Meadow shows that gesture can develop into language if it
assumes the full burden of communication (chapter 17), as it does in
the case of her young deaf subjects and also in the acquisition of a
conventional sign language (chapter 4). Gesture serves a quite
different function in oral languages, namely the imagistic
function. In every language both functions - the systemic and the
imagistic function - have to be covered. Usually, this happens in a
division of labor between the modalities: in sign languages the oral
modality is used for gesturing, in spoken language it is the gestural
modality. Susan Goldin-Meadow has also explored into this division of
labor and has found that in spoken language acquisition the
visual-gestural channel can provide valuable insights into the
developmental state of a child acquiring a spoken language. Sometimes
the child expresses knowledge in the gestural channel which it cannot
yet expressed in the oral channel, as in the case of gesture-speech
"mismatch". Gesturing is therefore helpful in the overall cognitive
development. In her book "Hearing gesture: How our hands help us
think" (2003b) Susan Goldin-Meadow gives a comprehensive account of
the diverse roles that gesture assumes in spoken language. Both
monographs supplement each other in an obvious way and witness the
deep involvement of the author in the interdisciplinary field of
gesture research.

It is the merit of Susan Goldin-Meadow's book(s) to focus our thinking
on the deep and still unresolved puzzles of language acquisition by
showing us what the very foundations of language are. Her
preoccupation with these fundamental questions for the last 25 years
presents us with a highly unique, original, and comprehensive treatise
in the field of language acquisition and cognitive development.

The only drop of bitterness I had to swallow in reading this
insightful and thought-provoking book concerns the fate of the young
subjects. While Susan Goldin-Meadow as the objective scientist is
silent on any ethical or moral implications, the burning issue to me
is: Were these fine young children eventually redeemed from their
deprivation and allowed to immerse into a full language model which,
no doubt, ideally would be a sign language? The deaf children did the
best they could - now it is up to their environment to provide them
with what is a human right: to have a full natural language.

REFERENCES

Bickerton, D. (1999): Creole languages, the language bioprogram
hypothesis, and language acquisition. In W.C. Ritchie and T.K. Bhatia
(eds.), Handbook of child language and acquisition (pp. 195-220). New
York: Academic.

Bierwisch, M. (2001): Repertoires of primitive elements. Prerequisite
or result of acquisition? In: J. Weissenborn & B. Hoehle (Eds.),
Approaches to bootstrapping. Phonological, lexical, syntactic and
neurophysiological aspects of early language acquisition. Vol. 2,
281-307. Amsterdam, Philadelphia: John Benjamins.

Elman, J., Bates, E., Johnson, M, Karmiloff-Smith, A., Parisi, D. and
Plunkett, K. (1996): Rethinking innateness. A connectionist
perspective on development. Cambridge, MA: MIT Press.

Goldin-Meadow, S. (2003a): Thought before language: Do we think
ergative? In D. Gentner & S. Goldin-Meadow (eds.), Language in mind:
Advances in the study of language and thought. Cambridge: MIT Press.
Goldin-Meadow, S. (2003b): Hearing gesture: How our hands help us
think. Cambridge, MA: Harvard University Press.

Hohenberger, A. (2002): Functional categories in language acquisition:
Self-organization of a dynamical system. Linguistische Arbeiten 456.
T�bingen: Niemeyer.

Jovanovic, B., Kir�lyi, I., Elsner, B, Gergely, G., Prinz, W. &
Aschersleben, G. (submitted): The role of effects for infants'
perception of action goals.

Karmiloff-Smith, A. (1992): Beyond modularity. A developmental
perspective on cognitive science. Cambridge, MA: MIT Press.

Prinz, W. (1997): Perception and action planning. European Journal of
Cognitive Psychology 9, 129-154.

ABOUT THE REVIEWER

My research interests are (i) first language acquisition, where I put
forward a dynamical account of the acquisition of syntax (ii) language
processing, where I compared slips of tongue and hand (iii) sign
language, especially processing and acquisition, and (iv) the
development of cognition and action as a broader framework relating
(i-iii) to human action.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue