CURRENT DIRECTIONS IN PSYCHOLOGICAL S CIENCE

Talking and Thinking With Our

Hands
Susan Goldin-Meadow

University of Chicago

ABSTRACT—When people talk, they gesture. Typically, where they are used (outside), and where they are kept (down-
gesture is produced along with speech and forms a fully stairs). For this child, the burden of communication has fallen on
integrated system with that speech. However, under unu- gesture and his gestures have risen to the occasion, assuming not
sual circumstances, gesture can be produced on its own, only the function of language but also many of its formal features,
without speech. In these instances, gesture must take over such as segmentation (producing separate gestures to represent
the full burden of communication usually shared by the two objects and the relations among them), combination (combining
modalities. What happens to gesture in this very different those gestures in a structured manner), and recursion (producing
context? One possibility is that there are no differences in more than one proposition within a single gesture sentence;
the forms gesture takes with speech and without it—that Goldin-Meadow, 2003a).
gesture is gesture no matter what its function. But that is The gestures that deaf children produce in place of speech
not what we find. When gesture is produced on its own and stand in sharp contrast to the gestures that hearing speakers
assumes the full burden of communication, it takes on a produce along with speech. Gestures that accompany speech
language-like form. In contrast, when gesture is produced share the burden of communication with that speech and, in-
in conjunction with speech and shares the burden of com- terestingly, do not assume a language-like form (McNeill, 1992).
munication with that speech, it takes on an unsegmented, These gestures are picture-like in form and rarely combine with
imagistic form, often conveying information not found one another to create sentence-like gesture strings. Neverthe-
in speech. As such, gesture sheds light on how people think less, gestures produced with speech are not mere hand-waving.
and can even play a role in changing those thoughts. They convey substantive information in their own right and may
Gesture can thus be part of language or it can itself be offer unique insight into a speaker’s unspoken thoughts (Goldin-
language, altering its form to fit its function. Meadow, 2003b).
KEYWORDS—gesture; sign language; cognitive load; com-
My goal in this article is to explore gestures of both types:
munication; instruction gestures that turn into language and reveal the basic capacity
we have for structured communication, and gestures that work
alongside language and shed light on how we think.
Imagine a deaf child whose hearing losses prevent him from
acquiring spoken language and whose hearing parents have
chosen not to expose him to a signed language. The child is, in WHEN GESTURE BECOMES LANGUAGE
effect, deprived of a model for language. We might expect such a
The Resilient Properties of Language
child to be unable to communicate. But we would be wrong.
When deaf children are exposed to sign language from birth,
Children in such circumstances do communicate: They gesture.
they learn that language as naturally as hearing children learn
For example, when shown a picture of a shovel, one such deaf
spoken language. However, 90% of deaf children are not born to
child produced iconic gestures for dig, snow-falls, and pull-on-
deaf parents who could provide early access to sign language.
boots and pointed outside and downstairs, thus conveying sev-
Instead, they are born to hearing parents who often choose to
eral propositions about snow shovels—how they are used (to
expose their children solely to speech. Unfortunately, it is un-
dig), when they are used (when it snows and boots are worn),
common for deaf children with profound hearing losses to ac-
quire spoken language, even with specialized instruction.
My colleagues and I have studied 10 profoundly deaf children
Address correspondence to Susan Goldin-Meadow, University of
Chicago, Department of Psychology, 5730 S. Woodlawn Avenue, in the United States and 4 in Taiwan. The children’s hearing
Chicago, IL 60637; e-mail: [email protected]. parents had decided to educate them in oral schools where sign

34 Copyright r 2006 Association for Psychological Science Volume 15—Number 1

 Susan Goldin-Meadow

TABLE 1
The Resilient Properties of Language as Manifested in the Gesture Systems of Deaf Children

Language property In deaf children’s gesture

Words
Stability Gesture forms are stable and do not change capriciously with changing situations.
Paradigms Gestures consist of smaller parts that can be recombined to produce new gestures with different meanings.
Categories The parts of gestures are composed of a limited set of forms, each associated with a particular meaning.
Arbitrariness The relation between gesture form and meaning, although essentially transparent (i.e., it is easy to guess
the meaning from the form), has arbitrary aspects.
Grammatical functions Gestures are differentiated by the noun, verb, and adjective grammatical functions they serve.
Sentences
Underlying frames Frames organized around the act predicate underlie gesture sentences.
Deletion Consistent production and deletion of gestures within a sentence mark particular thematic roles.
Word order Consistent orderings of gestures within a sentence mark particular thematic roles.
Inflections Consistent inflections on gestures mark particular thematic roles.
Recursion Complex gesture sentences are created by recursion.
Redundancy reduction Gestures are produced for redundant semantic elements in complex sentences less often than for
nonredundant semantic elements.
Language use
Here-and-now talk Gesturing is used to make requests, comments, and queries about the present.
Displaced talk Gesturing is used to communicate about past, future, and hypothetical events.
Narrative Gesturing is used to tell stories about self and others.
Self-talk Gesturing is used to communicate with oneself.
Generic statements Gesturing is used to make generic statements, particularly about animate objects.
Meta-language Gesturing is used to refer to one’s own and others’ gestures.

language was neither taught nor encouraged. The children had We found that the hearing mothers’ gestures did not have lan-
made little progress in oral language and, in addition, had not guage-like structure (e.g., Goldin-Meadow & Mylander, 1998).
been exposed to sign language. The children thus knew neither Thus, the children received as input speech-accompanying
sign nor speech. gestures that were not language-like in form, but produced as
Nevertheless, these children spontaneously used gestures to output gestures that resembled language.
communicate. What is particularly surprising is that the chil- Why didn’t the resilient properties of language appear in the
dren’s gestures displayed many of the structural properties of hearing mothers’ gestures? The mothers wanted their deaf
natural language. We have called the linguistic properties that children to learn to talk and, as a result, always spoke as they
the deaf children introduced into their gesture systems resilient gestured. We hypothesized that the mothers’ gestures (like the
properties of language (Table 1). The example at the beginning of gestures of all hearing speakers; Kendon, 1980; McNeill, 1992)
this article illustrates two such properties: recursion (the child were integrated with the words they accompanied and thus were
has expressed several propositions, each dealing with snow not free to assume the language-like properties found in their
shovels, within a single gesture sentence) and displaced com- children’s gestures. This hypothesis leads to the following pre-
munication (the child has described events that are not taking diction: Adults’ gestures should look more like those of the deaf
place in the here and now). children if they are produced without talking. We tested this
prediction experimentally.
Gesture In, Language Out
The deaf children in our studies were not exposed to sign lan- Turning Gesturers Into Signers
guage. They were, however, exposed to the gestures that their We asked English-speakers who had no experience with sign
hearing parents produced as they spoke. These gestures could language to describe videotaped scenes using their hands and
have served as input to the children’s gesture systems. To explore not their mouths. We then compared the resulting gestures to
this possibility, we looked at the gestures that the hearing gestures these same adults produced when asked to describe the
mothers produced when talking to their deaf children. However, scenes using speech (Goldin-Meadow, McNeill, & Singleton,
we looked at them not as they were meant to be experienced (i.e., 1996). When using gesture with speech, the adults rarely com-
with speech), but as a deaf child would look at them: We turned bined gestures into strings, and when they did, those gestures
off the sound and analyzed the mothers’ gestures using the same were not consistently ordered (Fig. 1A). In contrast, when using
analytic tools that we used to describe the children’s gestures. gesture on its own, the adults often combined gestures into

Volume 15—Number 1 35
 Talking and Thinking With Our Hands

Fig. 1. Examples of the gestures adults produce with and without speech. The top picture displays the event the adult is describing—a donut-
shaped object arcing out of an ashtray. The middle row of pictures displays the gestures the adult produces when asked to describe this event in
speech. Note that, although he produces several gestures, those gestures are separated by pauses (he relaxes his hands in his lap) and thus do not
form a single gesture string. In addition, the hand shapes he uses are loosely formed and sloppy. The bottom row of pictures displays the gestures
the adult produces when asked to describe the event using only his hands and no words. He now produces a string of gestures without breaking his
flow of movement, and his hand shapes become crisp and clearly articulated. In addition, his gestures adhere to a consistent order: stationary
object (ashtray) ! moving object (donut) ! action (arc-out). Reprinted from Hearing Gesture: How Our Hands Help Us Think, by Susan Goldin-
Meadow, 2003, Harvard University Press, pp. 234–235. Copyright 2003 by Harvard University Press. Reprinted with permission.

strings characterized by order and, interestingly, this order did forms of representation, the two modalities rarely contribute
not follow canonical English word order (Fig. 1B). identical information to a message.
To summarize thus far: When gesture is called upon to fulfill Nonetheless, the information conveyed in gesture and speech
all of the communicative functions of speech, it takes on the can overlap to a greater or lesser degree. For example, consider a
properties of segmentation and combination characteristic of hearing child asked first whether two identical rows of checkers
speech. This transformation happens in deaf children not ex- have the same number, and then whether the number of checkers
posed to a linguistic model and also in hearing adults asked on in one of the rows changes after the checkers are spread out. The
the spot to communicate only with their hands. The appearance child says that the number of checkers is the same at the be-
of these properties is particularly striking given that they are not ginning but different after the checkers are moved. When asked
found in the gestures that speakers routinely produce when they to explain this answer, the child mentions the movements the
talk. experimenter made when spreading the checkers apart in her
speech and gesture: She says ‘‘It’s different because you moved
WHEN GESTURE ACCOMPANIES LANGUAGE them’’ while producing a ‘‘spreading out’’ motion with her hands.
The child thus conveys a justification in gesture that overlaps a
Gesture Can Tell Us What Is on a Speaker’s Mind great deal with the justification in speech; she has produced a
(Even If She Doesn’t Say It) gesture–speech match.
Speech conveys meaning discretely, relying on codified words Now consider another child who gives the same explanation as
and grammatical devices. Gesture that accompanies speech the first child in speech, but conveys different information in
conveys meaning holistically, relying on visual and mimetic gesture: He moves a pointing finger from the first checker in row
imagery. Because gesture and speech employ such different 1 to the first checker in row 2 and then does the same with the

36 Volume 15—Number 1
 Susan Goldin-Meadow

TABLE 2
Examples of Matching and Mismatching Gestures on Mathematical Equivalence Problems

Math problem Speech Matching gesture Mismatching gesture

(a) 7 1 6 1 4 5 — 1 4 ‘‘I added the 7, 6, and 4 and got 17’’ Point at 7, 6, left 4, and the blank Point at 7, 6, left 4, right 4, and the
(add-to-equal-sign) (add-to-equal-sign) blank (add-all-numbers)
(b) 3 1 5 1 2 5 3 1 — ‘‘I added the 3, 5, 2, and 3 and got 13’’ Point at left 3, 5, 2, right 3, and V-shaped hand placed under the 5
(add-all-numbers) the blank (add-all-numbers) and 2, point at blank (group-and-
add-two-numbers)
(c) 6 1 3 1 4 5 — 1 4 ‘‘I made both sides of the problem Sweep under the left side of the Point at 6, 3, left 4, flick-away
equal’’ (equalize-two-sides) equation; sweep under the right movement near the right 4 (add-
side of the equation (equalize- numbers-on-left-then-subtract-
two-sides) number-on-right)

Note. Both of the strategies illustrated in problem (a) lead to an incorrect solution; the mismatching response thus contains two incorrect strategies. In contrast, in
problem (b), the strategy conveyed in the mismatching gesture leads to a correct solution, whereas the strategy conveyed in speech and the matching gesture leads to
an incorrect solution; the mismatching response thus contains a correct strategy (in gesture) and an incorrect strategy (in speech). In problem (c), both strategies
lead to correct solutions; the mismatching response thus contains two correct strategies.

second checkers in each row. This child focused on the exper- Teachers spontaneously give children who produce mis-
imenter’s actions in speech, but on the one-to-one correspond- matches on a task different instruction on that task than they give
ence between the checkers in the two rows in his gestures; he has children who produce only matches (Goldin-Meadow & Singer,
produced a gesture–speech mismatch. The child expressed an 2003). For example, teachers use more mismatches of their own
idea in his gestures that could not be found in his speech. (see example in Table 2, problem c) when teaching mismatchers
Children who produce gesture–speech mismatches on a task than they do when teaching matchers––and mismatchers learn
may have information relevant to solving the task literally at faster than matchers do.
their fingertips and could, as a result, be on the cusp of learning But do mismatchers learn faster because of the different
how to solve that task. If so, they ought to be particularly re- instruction they receive or because they are ready to learn? To
ceptive to instruction—and indeed they are. We asked children find out, we experimentally manipulated children’s instruction.
to solve a task and explain how they did it; none of the children Following a script, an experimenter taught children a correct
solved the problem correctly. During their explanations, some of strategy for solving a math problem in speech and gave some
the children produced gestures conveying the same information children the same correct (matching) strategy in gesture, some a
as their speech (matchers). Other children produced gestures correct but different (mismatching) strategy in gesture, and some
conveying different information from their speech (mismatch- no gesture at all (see Table 2, problem c). Children who were
ers). We then gave all of the children instruction in the task, and taught with mismatching gestures were more successful after
found that children who were mismatchers prior to instruction instruction than were children taught with matching gestures or
were more likely to profit from that instruction than children who no gestures (Singer & Goldin-Meadow, 2005).
were matchers (Goldin-Meadow, 2003b; see also Pine, Lufkin, & A conversation in gesture thus appears to be taking place
Messer, 2004). Interestingly, the mismatchers benefited from alongside the conversation in speech whenever speakers use
instruction even if both their gestures and their speech conveyed their hands. Children’s gestures reveal their cognitive state to
incorrect information (e.g., Table 2, problem a). their listeners who, in turn, provide instruction in gesture that
The gestures that children produce as they explain a task promotes learning. Learners use their hands to change their
reflect their knowledge of the task (see Goldin-Meadow, 1997). learning environments.
But current work suggests that gesture does more: It can play a
role in changing knowledge.
Our Gestures Can Change How We Think
Gesture also has the potential to function as a mechanism of
Our Gestures Can Change How Others React to Us change through its internal cognitive effects. When faced with
Gesture has the potential to function as a mechanism of change a difficult problem to solve, people find it helpful to externalize
through its communicative effects: (a) Speakers reveal infor- their thoughts—for example, writing a problem down, thereby
mation about their cognitive status through their gestures; (b) freeing cognitive resources that can then be used to solve the
listeners may pay attention to those gestures and alter their input problem. Can gesture, like writing, improve learning by influ-
accordingly; (c) speakers could then profit from this altered in- encing learners directly?
put. We have just reviewed evidence for the first step in this Children imitate the gestures that a teacher produces. More-
process. The evidence for the second and third steps comes from over, children who produce these gestures are more likely to
math tutorials involving teachers and individual children. succeed after instruction than are children who do not (Cook &

Volume 15—Number 1 37
 Talking and Thinking With Our Hands

1.0 should not matter; it should only matter that a speaker gestures,
No Gesture by Instruction
No Gesture by Choice not what the speaker gestures. But the number of items that
* Gesture speakers remember does depend on the meaning conveyed by
Proportion of Items Recalled Correctly

0.8
gesture. Speakers remember more items when their gestures
convey the same information as their speech (i.e., one piece of
*
information about the problem) than when their gestures convey
0.6
different information (two pieces of information; Wagner et al.,
2004). Gesture’s content thus determines demands on working
0.4
memory, suggesting that gesture confers its benefits, at least in
part, through its representational properties.

0.2 WHERE TO NEXT?

Gesture is chameleon-like in form, and that form is tied to

0.0
Verbal Visuospatial
function. When gesture assumes the full burden of communi-
Type of Memory Load cation, acting on its own without speech, it takes on language-
*p < .05, Gesture vs. No Gesture by Instruction; Gesture vs. No Gesture by Choice like form (as in the deaf children described earlier and the
Fig. 2. Proportion of verbal and visuospatial items that adults recalled
hearing adult in Fig. 1B). But when gesture shares with speech
while explaining how they solved a math problem. Adults remembered the burden of communication, it loses its language- structure and
more items of both types when they spontaneously gestured during their assumes instead a holistic form (as in all hearing speakers, in-
explanations than when they did not gesture—either when they were in-
cluding the deaf children’s hearing mothers and the adult in Fig.
structed not to gesture or when they spontaneously chose not to gesture.
Gesturing thus frees up cognitive resources that can then be used on either 1A). Although not language-like in structure, gesture that ac-
a verbal or visuospatial task. Adapted from Wagner, Nusbaum, and Gol- companies speech is not just handwaving. It conveys information
din-Meadow (2004, Fig. 3).
imagistically and, as such, accesses different information than
speech does. Gesture thus lets speakers convey thoughts they do
Goldin-Meadow, in press). Thus gesturing during instruction not have words for and may even play a role in changing those
might be effective because it encourages children to produce thoughts. But many questions remain.
gestures of their own, which, in turn, leads to learning. Learners First, the left hemisphere is specialized for processing lin-
use their hands to change their minds. guistic information, be it spoken or signed. Do the gestures
But how? It may be that gesturing lightens cognitive load. Adults hearing speakers produce with speech show left-hemisphere
and children were asked to explain how they solved a math problem dominance? We do not know the answer yet, but it is likely to be
while at the same time remembering a list of words or letters. Both no, as these gestures do not exhibit the hierarchically segmented
groups remembered more items when they gestured during their structures found in speech and sign. Do the language-like ges-
math explanations than they did when they did not gesture (Goldin- tures deaf children of hearing parents use instead of speech show
Meadow, 2003b). Gesturing appears to save speakers cognitive left-hemisphere dominance? Again we do not know, but the
resources on the explanation task, permitting them to allocate more answer is likely to be yes, as these homemade gesture systems, if
resources elsewhere—in this case, to a memory task. truly linguistic, ought to be processed like natural language.
But gesture might not be lightening the speaker’s load. It might Second, do adults and children differ in how they use gesture
be merely shifting the load from a verbal memory store to a with speech and without it? Although adults can introduce many
visuospatial memory store. The idea here is that gesturing allows linguistic properties into gesture when asked to use it without
speakers to convey in gesture information that might otherwise speech, their gestures lack certain properties found in the deaf
have gone into a verbal store. Lightening the burden on the children’s gestures (Singleton, Morford, & Goldin-Meadow, 1993).
verbal store should make it easier to do a verbal task that is In contrast, adults and children seem to use gesture with speech
performed simultaneously. If, however, the burden has really in the same ways. But do these gestures serve the same functions
been shifted to a visuospatial store, it should be harder to per- for adults and children?
form a spatial task (such as recalling the location of dots on a Finally, how does cultural variation affect gesture produced
grid) when simultaneously gesturing than when not gesturing. with speech and without it? American and Chinese deaf children
But we have found that gesturing continues to lighten the invent gesture systems with many similarities despite differ-
speaker’s load even if the second task is a spatial one (Fig. 2; ences in their hearing parents’ gestures. What about deaf chil-
Wagner, Nusbaum, & Goldin-Meadow, 2004). dren in other cultures (e.g., in Nicaragua, where deaf children
Perhaps gesturing lightens a speaker’s load because it is a were brought together in a group for the first time in 1980, al-
motor activity that energizes the memory system (Butterworth & lowing gesture creation to take place within a social community;
Hadar, 1989). If that were the case, the type of gesture produced Senghas & Coppola, 2001)? And does gesture produced with

38 Volume 15—Number 1
 Susan Goldin-Meadow

speech play the same role in thinking for hearing speakers Goldin-Meadow, S. (1997). When gesture and words speak differently.
around the globe? Current Directions in Psychological Science, 6, 138–143.
Gesture offers insight into the basic capacity we have for Goldin-Meadow, S. (2003a). The resilience of language: What gesture
creation in deaf children can tell us about language-learning in
structured communication when produced without speech and
general. New York: Psychology Press.
into how we think when produced with speech. The time is ripe Goldin-Meadow, S. (2003b). Hearing gesture: How our hands help us
to take advantage of our hands. think. Cambridge, MA: Harvard University Press.
Goldin-Meadow, S., McNeill, D., & Singleton, J. (1996). Silence is
liberating: Removing the handcuffs on grammatical expression in
Recommended Reading the manual modality. Psychological Review, 103, 34–55.
Beattie, G. (2003). Visible thought: The new psychology of body lan- Goldin-Meadow, S. & Mylander, C. (1998). Spontaneous sign systems
guage. New York: Routledge. created by deaf children in two cultures. Nature, 91, 279–281.
Kendon, A. (2004). Gesture: Visible action as utterance. Cambridge, Goldin-Meadow, S., & Singer, M.A. (2003). From children’s hands to
England: Cambridge University Press. adults’ ears: Gesture’s role in teaching and learning. Develop-
Liddell, S.K. (2003). Grammar, gesture and meaning in American Sign mental Psychology, 39, 509–520.
Language. Cambridge, England: Cambridge University Press. Kendon, A. (1980). Gesticulation and speech: Two aspects of the
McNeill, D. (Ed.) (2000). Language and gesture. Cambridge, England: process of utterance. In M.R. Key (Ed.), Relationship of verbal and
Cambridge University Press. nonverbal communication (pp. 207–228). The Hague: Mouton.
McNeill, D. (2005). Gesture and thought. Chicago: University of Chi- McNeill, D. (1992). Hand and mind. Chicago: University of Chicago
cago Press. Press.
Pine, K.J., Lufkin, N., & Messer, D. (2004). More gestures than answers:
Children learning about balance. Developmental Psychology, 40,
1059–1067.
Acknowledgments—This research was supported by grants Senghas, A., & Coppola, M. (2001). Children creating language: How
from the National Institute on Deafness and Other Communi- Nicaraguan Sign Language acquired a spatial grammar. Psycho-
cation Disorders (R01 DC00491) and the National Institute of logical Science, 12, 323–328.
Singer, M.A., & Goldin-Meadow, S. (2005). Children learn when their
Child Health and Human Development (R01 HD47450).
teacher’s gestures and speech differ. Psychological Science, 16,
85–89.
REFERENCES Singleton, J.L., Morford, J.P., & Goldin-Meadow, S. (1993). Once is
not enough: Standards of well-formedness in manual communi-
Butterworth, B., & Hadar, U. (1989). Gesture, speech, and computational cation created over three different timespans. Language, 69,
stages: A reply to McNeill. Psychological Review, 96, 168–174. 683–715.
Cook, S.W., & Goldin-Meadow, S. (in press). The role of gesture in Wagner, S.M., Nusbaum, H., & Goldin-Meadow, S. (2004). Probing the
learning: Do children use their hands to change their minds? mental representation of gesture: Is handwaving spatial? Journal
Journal of Cognition and Development. of Memory & Language, 50, 395–407.

Volume 15—Number 1 39