Full Research Report
Title: Phonological properties of hand shapes in home signs used by deaf children in
hearing family
Grant reference: ES/J00362X/1
Sotaro Kita (PI)
Collaborators: Susan Goldin-Meadow, Diane Brentari
RAs: Yousaf Kanan, Beinan Zhou
Introduction
Children across the globe learn, with seeming ease, whatever language they
are exposed to. It is clear that the linguistic input that children receive from the
environment has a significant impact on language development. After all, children
who grow up in an English-speaking environment learn specific features of English
that other languages do not necessarily have. However, some properties of language
seem to be developmentally resilient in the sense that children will develop them even
if the input provides no (or highly degraded) information about them in situations
such as congenitally deaf children who are not exposed to sign language (GoldinMeadow, 2003). In other words, children seem to be born with the ability to develop
certain properties of language. Exactly what properties of language are
developmentally resilient is an important question in developmental psychology and
biology. This is because it can shed light on the "nature-nurture debate" on language
development by specifying how biological endowment of children contributes to
language development. The proposed project will examine the resilience of language
in the domain of phonology.
The most direct way of investigating developmentally resilient properties of
language is to look at a situation in which linguistic input is highly degraded or absent
and see what properties of language develop despite the poverty of input. Such a
situation arises in congenitally (profoundly) deaf children who grow up among
speakers of spoken language and do not get exposure to conventional sign language.
Idiosyncratic gestural communication systems developed by congenitally deaf
child growing up with hearing family are called "home sign" (Goldin-Meadow, 2003;
Goldin-Meadow & Feldman, 1977; Goldin-Meadow & Mylander, 1983). Home sign
systems have properties that resemble properties of conventional signed and spoken
languages. Furthermore, such properties are not observed in speech-accompanying
gestures (McNeill, 1992) produced by the parents. These properties can be seen as
resilient properties of language because they emerged from deaf children despite the
fact that they were not available in the input. The previous research on home signs
demonstrated that certain aspects of syntax and morphology are resilient against the
poverty of input.
Objectives
The aim of the project was to investigate whether or not "home signs"
created by deaf children without linguistic input (to gesturally communicate
with family members) had the foundation of phonological systems that are seen
in conventional spoken and signed languages, namely "discreteness" and "duality
of patterning" (i.e. existence of discrete categories of forms that are independent
from semantic categories, e.g. phonological features). To this end, we compared
types of hand shapes used in deaf children's home signs and their hearing
parents' gestures, based on existing recordings of interaction between deaf
children and their parents while playing with toys. Hand shapes are considered
to be phonological elements in sign language. In order to assess cross-cultural
generalisability, we investigated home sign systems in the USA and in Taiwan.
Two analyses were carried out. First, we tested if home signs created by deaf
children had more discrete (as opposed to continuous) handshape categories, as
compared to gestures used by the deaf children's parents. Second, we tested if
the discreteness of the handshape varied between two different types of gestures
(pointing vs. conventionalised gestures) with different meanings/functions.
Methods
Participants and recordings
We used the existing recordings of congenitally deaf children (four in the
USA and four in Taiwan) interacting with their mother with a standard set of
toys. Each child was observed twice between 3 years 8 months and 4 years 11
months (each session lasted one to two hours). This is the dataset used in
Goldin-Meadow and Mylander (1998). These children had 70 to 90 dB hearing
loss in both ears. They attended oral-education schools and were trained in lipreading and speech production. But at the point of recording, none had acquired
spoken language beyond an occasional isolated word. None of the children were
exposed to conventional sign language. The children's hearing parents tried to
communicate with the children in spoken language. However, much of the
communication between the children and the parents occurred via gestures and
actions.
Hand shape coding
The first goal of the coding was to capture the extent to which gestures
have discrete categories of hand shapes as opposed to hand shapes varying along
a continuum (e.g. a continuum between a fully extended or fully curled-in middle
finger as in Figure 1). To this end, we coded whether observed hand shapes fall
between a "grid" of canonical hand shapes for phonetic transcription for hand
shape in sign language (Eccarius & Brentari, 2008). We classified the hand shape
that fell between the grid as "intermediate". The grid was detailed enough to
capture most of the hand shapes that occur in various sign languages. The hand
shapes in Figure 1a and 1c are in the grid, but the hand shape in Figure 1b is not.
The hand shape is evaluated at the beginning and the end of the meaning
bearing part of the gestural hand movement, called "gesture stroke" (Kita, Gijn, &
Hulst, 1998). More specifically, we coded (a) to which hand shape in the
Eccarius-Brentari chart it resembled most, and (b) if it was a canonical or
intermediate hand shape (see Figure 1).
(a) canonical
(b) intermediate (c) canonical
Figure 1. Examples of canonical hand shapes (a, c), in which fingers are clearly
selected or unselected, and an intermediate hand shape (b), in which some
fingers (in this example the middle finger) are somewhere between selected and
unselected.
Furthermore, for each gesture, we coded whether finger selection in the
hand shape is crucial for the meaning of the gesture or not. For example, if a
gesture with the extended index finger hand shape (Figure 1a, 1b) was used to
point to an object, it was coded as "finger selection not crucial" because pointing
can be achieved by any number of fingers extended. In contrast, if a gesture
refers to a thin elongated object (e.g. a stick), it was coded as "finger selection
crucial" because the hand shape was iconically related to the meaning.
Coding was carried out by using the video annotation software, ELAN
<www.mpi.nl/tools/elan>.
Results and Discussions
There was some evidence that deaf children introduced "discreteness" of
form units in their communication systems. We analyzed what proportion of
gesture tokens had a "canonical" handshape in deaf children and their mothers
and in the US and in Taiwan. When canonical handshapes were observed more
often, the communication system was considered to be more discrete (i.e.,
handshapes clustered around canonical targets). For the US samples, the deaf
children's home signs were more likely to have a canonical handshape
(proportion of canonical gesture/sign tokens: M = .79) than their mothers'
gestures (M = .73) (p < .05). However, for the Taiwanese sample, children's
home signs (M = .67) and parents' gestures (M = .68) did not significantly differ
from each other.
This cultural difference is consistent with the previous analyses of the
same set of recordings. Goldin-Meadow and Mylander (1999) investigated
complex gesture combinations expressing events (e.g., a sequence of three
gestures to express the event, a mouse eating cheese) in deaf children's home
signs and mothers' gestures. They found that, in the Taiwanese sample, deaf
children's and their mothers' gesturing patterns tended to converge with each
other, unlike in the US sample. We speculate that this may be because of culturespecific attitude towards the role of gesturing in communication.
The results showed evidence for duality of patterning in deaf children's
home signs for the US samples, but not for Taiwanese sample. Duality of
patterning refers to the idea that form units are organised independently of
meanings and functions. We investigated whether the discreteness of
handshapes were modulated by meaning/function of home signs and gestures.
More specifically, we compared discreteness of the handshapes for pointing
gestures and conventionalised gestures (such as the gesture that showed a palm
of the open hand to the recipient to express the command, "stop"). If form units
are organised independently of meaning, the degree of discreteness should not
differ between different types of gestures. Alternatively, if form units are shaped
by function/meaning, then pointing should be less discrete than
conventionalised gestures because discreteness is less important for the function
of pointing than that of conventionalised gestures. For the US samples, this
duality of patterning was found in deaf children's home signs (proportion of
canonical gesture/sign tokens: pointing, M = .83; conventionalised, M = .82), but
not in the mothers' gestures (pointing, M = .83; conventionalised, M = .74;
pointing > conventionalised, p = .009) (interaction between age and gesture type,
p = .026). For the Taiwanese sample, we found no evidence that duality of
patterning differ between children's home signs (pointing, M = .65;
conventionalised, M = .79) and mothers' gesture (pointing, M = .70,
conventionalised, M = .80) (no significant main effects or interaction).
To summarise, we found some evidence that deaf children spontaneously
introduce discreteness and duality of patterning, two foundational properties of
language, into their home signs. This pattern was seen in the US sample, but not
in the ROC sample. The cultural difference is an important topic for future
research.
References
Eccarius, P., & Brentari, D. (2008). Handshape coding made easier: A theoretically based
notation for phonological transcription. Sign Language and Linguistics, 11(1), 69101.
Goldin-Meadow, S. (2003). The resilience of language: What gesture creation in
deaf children can tell us about how all childrenlearn language. New York:
Psychology Press.
Goldin-Meadow, S., & Feldman, H. (1977). Development of language-like
communication without a language model. Science, 197(4301), 401-403.
Goldin-Meadow, S., & Mylander, C. (1983). Gestural Communication in DeafChildren - Noneffect of Parental Input on Language-Development. Science,
221(4608), 372-374.
Goldin-Meadow, S., & Mylander, C. (1998). Spontaneuos sign systems created by
deaf children in two cultures. Nature, 391, 279-281.
McNeill, D. (1992). Hand and mind. Chicago: University of Chicago Press.