topics: |  cognitive | blind | developmental | psychology | language |

Excerpts

The ways into language seem to rely heavily on exploiting the
non-linguistic and immediate environment, and on utilizing visually based
strategies for gaining and directing attention.  ix
`%
Visual information has been [related to] many facets of the process of
language acquisition as it normally progresses. ... thought to be important
in fostering early parent-infant interaction, in providing the child with a
stimulus for hypothesizing about what language encodes, and in supplying the
parents with clues about what a young child's early verbalizations
mean. Furthermore, vision seems to be crucial in the infant's
conceptualization of the environment, on which early language development is
thought to depend.

Visually based strategies

Stern (1974, 1977) points out that the infant's first exposure to the human
world is composed of his mother's activities, especially her repertoire of
"infant elicited behaviors."  These center on exaggerated facial
expressions, accompanied by vocalizations and gazing at the infant. The
human neonate has a strong propensity to observe and even imitate these
expressions (Meltzoff and Moore, 1977), with the result that they form the
core of interactive play.

The play episodes themselves are typically initiated by caregivers using a
combination of eye gaze and vocalization in which the objective is to
obtain mutual orientation in a face-to-face position with the infant
(Stern, 1977; Tronick et ai, 1979; Kaye, 1979). Once the infant is
attending, a play-dialogue ensues until the infant disengages by glancing
away. The best predictor of when a mother will respond again to the infant
is the moment that the infant's gaze again focuses on her (Brazelton et aL,
1974; Stern, 1974; Fogel, 1977).

Thus, visual attention on the part of both the caregiver and the infant is
crucial in initiating and maintaining early exchanges.  These rudimentary
exchanges are structured along the same lines as the adult discourse
system...  Without access to visual information, the structure of these
interactions is necessarily disrupted and there seems to be no substitute
for their effectiveness in establishing a bond between parents and infants,
and in initiating the human infant into the social world from which
language emerges (Fraiberg, 1977).

As the child begins to use language, visual information seems to provide an
important stimulus for building hypotheses about meaning. For example, in
ascribing meaning to words, the child appears to abstract certain salient
attributes from early referents and uses these as a basis for extending the
domain of application for words (Bowerman, 1976, 1978; Clark, 1973;
Nelson, 1973a; Rescorla, 1980). This process is essential in helping the child
move from using a word as a "name" for a specific referent, to using words as
symbolic vehicles to denote a heterogeneous class of referents.

Role of visual atttributes : Classifier systems

The overwhelming evidence is that such visual properties as shape,
size, and movement are the most important criteria used in constructing these
classifications. Not only is visual information important in the child's
organization of referent properties, but it appears to underlie adult
categorization and the structure of many lexical fields as well (Andersen,
1978; Clark, 1977b; Rosch, 1975, 1977).

Some of the evidence for this comes from the analysis of classifier systems
in a variety of natural languages.  Classifiers are expressions which group
together entities that share some particular attribute. English does not
exploit these, though the principle can be seen in the utterance "She
bought four lengths of material." Some languages classify all varieties of
countable objects yielding such sentences as "She has seven round-things
eggs."

Clark's (1977b) analysis of classifier systems reveals that perceptual
information, again largely visual, is the primary basis of groupings. The
features round, long and flat are especially important. For example, the
Indonesian language groups such objects as fruit, peas, eyes, balls, and
stones together on the basis of roundness; Nung groups together trees,
bamboo, thread, nails, and candles on the basis of length; Kachari groups
together leaves, fans, and cloth on the basis of flatness. Even in cultures
which do not have classifier systems, these features are important.

As the result of a number of experiments conducted in a variety of
cultures, Rosch (1973, 1975, 1977) has found that people tend to group
objects on the basis of perceptual features, especially the visual
perception of shape... defining lexical classes from infancy on, and visual
information seems to be central in this.

... The Piagetian notion of interaction with the environment as the basis
of sensorimotor intelligence specifically involves perception, especially
visual attention to objects and events, as well as purely motoric
behavior....    It appears that the course of development is hampered
for infants who are born blind (Piaget and Inhelder, 1969).

This is significant for the present discussion, since language acquisition
seems to depend on the emerging conceptual system and the young child's
developmental task involves matching linguistic and cognitive structures
(Clark, 1977a; Nelson, 1974; Pylyshyn, 1977). One area where this is
particularly evident is in the child's early expression of semantic roles
in which such fundamental relations as Agent + Object or Agent + Action are
thought to reflect the child's understanding of and experience with his
immediate environment.  Perhaps the most frequent explanation for how very
young children come to understand and produce language is that they depend
on the "here and now," which has been defined as "whatever is directly
under the child's eyes" (Clark and Clark, 1977, p. 322).

The child learns about the matching of language and world largely through
context.

Following others' gaze

A now classic example is Shatz's (1974) analysis of how toddlers
successfully respond to such directives as "Can you shut the door?" ... the
child follows his parent's eye gaze and gesture which are directed toward
the door, a strategy which crucially depends on vision. The child's
previous observations and explorations equip him with the knowledge that
doors can be opened and closed, and the child may pick up the parent's
intonation and recognize the utterance as a directive. p.4

Perception of distant objects (sun stars) large objects (mountain rivers)
etc. is more difficult.  No evidence that senses are finer - tactile, hearing
are both about the same.  However,
senses are used more effectively, e.g. echolocation - bright blind children
spontaneously learn to clap and echolocate around age 1.5-2.  10

congenitally blind, if vision is restored as adult by surgery, have
difficulty recognizing objects visually - need training.  [Molyneux problem] 11



Blind infants - not much work on testing them - devising elicitations is more
difficult.  11

First vocalizations: relatively stable sound pattern in particular
situation.  Differs from pre-lg babbling in that they are stable and repr a
consistent communicative function - but not actually words - may disagree
with adult usage.  Diary studies indicate that these patterns originate in
their imitating various environmental sounds, incl their own spontaneous
noises (Guillaume 78, Leopold 1939,1949;

Emergence of object concept

object concept related to ear-hand coordination (similar to eye-hand in
sighted child).
reaching for objects is a bit delayed - in sighted child around 4-5 mos, in
the blind, averaging 9 mos (6.5 to 11 mos).  [Warren 70, Bower 74]

[Blind children do not appear to have a clear distinction between spatial
deictives like here and there, this and that.

also slower to learn spatial concepts]

Blindness and childhood

Fraiberg (1977) used the Bayley Scales to document language acquisition in
ten blind subjects who were participating in her developmental/psychoanalytic
study of blindness in children. The Bayley items record only the very
broadest achievements (e.g., "says two words") and are limited to the very
earliest aspects of language. The results suggest language development in
blind children progresses at a normal pace, though different skills may have
ontogenetically differing significance for the two groups. For example,
Fraiberg's subjects are precocious in their attainment of the items "jabbers
expressively" and "imitates words" while they are delayed in "says two words"
and "sentences of two words". The first differences may indicate the
importance of acoustic signals to blind children, whereas later acquisition
of words and word combinations is in line with the reported delay in object
concept for blind children.

Language acquisition in blind children - point to differences:
 - Urwin (1978a),
 - Rowland (1980),
 - Mulford (1980, 1986),
 - Kekelis (1981), Mills (1983b),
 - Landau and Gleitman (1985),
 - Wilson (1985) and Bigelow (1987).

Urwin's pioneering study reveals that nursery rhymes incorporating
body actions, such as clapping or rocking, provide an unusually effective
means for parents to gain and maintain their blind infant's attention. These
routines may then become the basis of social play and provide a means for the
blind children to initiate and control interaction during their second
year. Interestingly, nursery routines became less important for one subject
Urwin studied who had some residual vision, a pattern more typical of sighted
children. Language itself seemed to evolve out of these rituals for her
subjects; a number of early words and phrases derived from familiar
routines. Nursery routines may thus provide an alternative to the visually
based pointing, gesturing and offering games that evolve in interactions with
sighted children and that seem to facilitate language (see Ninio and Bruner,
1978; Masur, 1982).

The impact of verbal routines may in fact be quite pervasive. Urwin suggests
that the beginnings of representational play in her blind subjects were not
in games with objects, but in reconstructing conversations.  Beginning at 1.5
years, one subject actually used different voice qualities to mark different
people's speech: for example, a gruff voice for his father and a lowvoicefor
his mother. The important point here isthat blind children may use language
to begin to engage in role-play and possibly to represent themselves. This
contrasts with Fraiberg's (1977) view that self representation (and correct
use of pronouns) evolves much later in blind children and may be associated
with object play.

Urwin also found that a rapid expansion in vocabulary at about 18 months
occurred concurrently for her children with the emergence of representational
play. This play involved objects for the child with residual vision but
mainly verbal role play for the totally blind children. However, ... a close
examination of roleplay speech reveals some difficulties blind children have
in understanding reversibility of perspective and deictic terms (Andersen,
Dunlea and Kekelis, 1984).

Just as verbal role-play may be characteristic of blind children, there is
also a tendency for them to engage in a considerable amount of sound and word
play not directed to others and there is a striking tendency to reproduce
segments of speech that sound like their caregiver's language (Urwin, 1978a;
Wilson, 1985; Fraiberg, 1977). This characteristic is what some clinicians
call "delayed echolalia" but it is important to distinguish between children
whose language is almost exclusively echoic and children who use language in
many situations including echoically. Only the former is regarded as
pathological. Moreover, sighted children also pick up and use unanalyzed
chunks of language which they eventually dissect (Peters, 1977, 1983).

... Mothers appear to compensate for their infant's visual impairment by
using greater physical contact and vocalizing to respond not only to their
infant's vocalizations, but also to their smiles and other behaviors (Urwin,
1978a; Rowland, 1980, 1984). Language inevitably serves an important function
in maintaining contact. Speech directed to blind children may be more
centered on the children themselves than on their activities with objects
(Urwin, 1978a; Kekelis, 1981; Andersen and Kekelis, 1982). In comparison with
sighted children, blind children seem to receive fewer statements describing
activities and events and more labels for objects or requests for the child
to identify objects. Kekelis suggests that this is because it is more
difficult to direct blind children's attention outside themselves and to
monitor their attention to events. In general, she suggests these strategies
appear to encourage visually impaired children to take an active role in
conversations, but they may also limit the kinds of information provided to
these children.

After a possible mild delay in beginning to say words, blind children seem to
acquire 50 or so words in the usual amount of time reported for sighted
children - a few months - then, as their vocabularies expand, they begin to
say two- and three-word phrases. Several researchers looking at the early
stages of language development have found blind and sighted children's
language fairly similar (Bigelow, 1986; Landau and Gleitman, 1985). The
strongest claim comes from Landau and Gleitman who argue that "the blind
learner can surmount whatever obstacles diminution of experience places in
her path, acquiring her native tongue in a largely unexceptional fashion"
(p. viii). They later suggest that blind and sighted children are
linguistically indistinguishable by 36 months. Such claims are based on
surveying the content ofearly vocabularies and the semantic roles expressed
in early multiword phrases and by calculating mean length of utterances.2

A number of researchers have found the content of blind children's early
vocabularies fairly similar to sighted children's when classified along the
lines proposed by Nelson (1973a) (Bigelow, 1986; Landau and Gleitman,
1985; Urwin, 1978a). In combining data from a number of sources,
including some unpublished data from the present study, Mulford (1986)
found the most notable differences were a relatively high proportion of
specific nominals and action words in the blind children's vocabularies, a
lower proportion of general nominals and surprisingly few function words
(e.g. "more," "what," "all gone").

There are fewer studies of blind children's early multi-word language.
Landau and Gleitman report that the three children they studied expressed
similar types of semantic roles to the sighted children studied by Bloom,
Lightbown and Hood (1975). [But] the vague category "other" is relatively
large for the blind children, leading the reader to wonder if some
interesting differences remain undiscovered.  (Landau and Gleitman, p. 37)

Urwin:  classified the two-word combinations produced by her oldest subject
according to Brown's (1973) semantic categories.
While all of the categories were attested, locative
relationships were rarely expressed and the utterances often referred to the
child's own activities. As we shall see later, this hints at an important
thematic difference in the way blind and sighted children use language.

The greatest delay reported at the multi-word level is Landau and Gleitman's
finding that their blind subjects were late in acquiring auxialiary
verbs. They attribute this to a proportionally greater number of declaratives
in mothers' language to their blind children and a correspondingly fewer
number of yes-no questions. In English, auxiliaries are stressed in yes-no
questions and are therefore relatively salient in these structures, which
would presumably aid children in learning them.

other areas in which blind children's
language development is seen as different from sighted children's. In
particular, Mills' discussion of phonology (1983a) and Mulford's work on
reference (1980, 1983) suggest some difficulties.
Mulford's investigation of referential terms in three to five year old blind
children indicates that pronoun use appears to be the same for her blind
subjects as for the sighted five year olds described by Wales
(1979). However, she did find the blind children making semantically based
errors.

* exophoric pronominal references, those which relate to experience and the
  situational context, were less frequent than endophoric
  references for the blind children.

* Endophoric references relate to language and the linguistic context, and
  may therefore be more accessible to the blind.

Perhaps most interesting, Mulford found that the blind children used deictic
terms such as "this" and "there" as names for locations and entities but were
not sensitive to the relative value of deictics with respect to the
speaker. Mulford's analysis demonstrates the importance of delving beneath
the surface to examine qualitative differences in use that may be obscured by
simple counts of arguments of the verb and the deictic referents that
co-occur with these interjectives. It would seem that this subject is using
an inferential strategy based on syntactic, semantic, and pragmatic
information and not simply the syntactic strategy Landau and Gleitman
propose.

First Words

The pioneering work of Clark (1973, 1975) and Nelson (1973b) assumes 
that children build word meanings on the basis ofone or two stable elements 
ofmeaning. Clark's theory ofsemantic feature acquisition, based on analysis 
of numerous diary records, proposes that children begin by associating only 
one or two features with a word and gradually add additional criterial 
features until the child's understanding of the term corresponds to adult 
usage. 

The child's first hypotheses about what words mean stem from the conceptual
organization of non-linguistic information, particularly the perceptually
based features of shape, movement, size, texture, sound, and taste. For any
given word, children initially command a limited number of features and may
overextend the word to cover other referents which share one or more of the
features. The following example based on shape is illustrative:

	Lexical item: mooi (child acquiring English) 
	First referent: moon 
	Domain of application: cakes, round marks on windows, 
			and in books, round shapes in 
			books, tooling on leather book 
			covers, round post marks, letter 'O' 
			Note: based on Clark (1973), p. 80 

There are also instances of "partial overextension" where a term is extended
to new referents which have just one of two dominant features (e.g., the
features of movement and roughness associated with a toy goat are the basis
of one child's extending his name for that toy to all moving objects and also
to all rough surfaces).

Although organizational strategies are thought to underlie early 
lexical development, many, psychologists hold that young children are 
incapable of classifying (see Inhelder and Piaget, 1964; Vygotsky, 1962), 
while others hold that intensive classification does originate during the 
sensorimotor period (Cohen and Strauss, 1979). 

Nelson's investigation of spontaneous sorting in infants between the ages of
12and 24 months indicates that children engage in consistent sorting and
grouping activities prior to the acquisition ofrelevant language and that the
sorting criteria are generally functional in nature, that is, they are based
on the child's understanding of how an item is used and of the dynamic states
of objects. In this view, objects are initially viewed in terms of their
"functional cores" from the child's perspective. Subsequent researchers have
interpreted Nelson's position to be that lexical meaning is based on the
extension of this functional information (see Barrett, 1978; Gentner, 1978;
Press, 1974; Thomson and Chapman, 1977). In fact, Nelson proposed that while
functional criteria are initially defining, extensions may also proceed on
the basis of perceptual criteria.

Feature component model vs Prototype Theory

The notion that children build word meanings on the basis of one or two
stable elements of meaning contrasts with the more traditional view in
psychology that early word meanings are used as "complexive" groupings.
That is, a word is extended on the basis of some recognized similarity
between two referents, then may be extended to a third referent on the
basis ofsome feature shared with a preceding referent, but the various
referents do not all share the same features. Vygotsky's (1962) classic
example is ofa child who extends the word "quah" from a duck swimming in a
pond, to all liquids, then to a coin with an eagle on it, and finally to
all round coin-like objects. In fact, there are reports of both categorical
and complexive uses of early words and neither strategy seems to have
ontogenetic priority over the other.

An alternative to the feature component model of lexical development draws
on the prototype model of category structure (see Rosch, 1973, 1975a, 1977;
Rosch and Mervis, 1975). The crucial insight is that categories for both
adults and children may consist of a core meaning, or a focal exemplar,
which is surrounded by other category members of progressively decreasing
similarity to the core meaning. Category boundaries are by definition
vague, since peripheral members may exhibit a fairly high degree of
fluidity: they may shift category membership in response to circumstances
prevailing at a given time. A good exemplar of the category "bird" is a
robin, while a penguin isa more peripheral member. An implication of this
theory for child development is that rather than focusing only on the
child's acquisition of criterial attributes, researchers should also
consider the role of core meaning in children's initial lexical
classifications.

Application of the theory reveals that the complexive strategies used in
early categorization are not necessarily the primitive "chained complexes"
proposed by Vygotsky that reflect unstable categories, but rather that
category membership isa matter ofdegree ofvariation from a prototype (see
Bowerman, 1976, 1977, 1978, 1979 and for slightly older children Andersen,
1975 and Anglin, 1977). All referents share something, but not necessarily
the same thing, with a prototype referent. This is of course similar to
Clark's notion of partial extension.