The Performance of Typically
Developing 21/2-Year-Olds on
Kathryn D. R. Drager
Janice C. Light
JoHannah Curran Speltz
Karen A. Fallon
Lauren Z. Jeffries
The Pennsylvania State University
University Park
298 Journal
Journal of of Speech,
Speech, Language,
Language, andand Hearing
Hearing
Research
Dynamic Display AAC Technologies
With Different System Layouts and
Language Organizations
The current generation of augmentative and alternative communication (AAC)
technologies is largely based on conceptual models of adults who are not
disabled (J. Light & P. Lindsay, 1991). As a result, there is a large “cost of
learning” placed on young children. This paper presents the results of a study
designed to investigate the learning demands of dynamic display systems that
differed in system layout and language organization for children approximately
21/2 years old (2 years 5 months to 2 years 11 months). Thirty typically develop-
ing children were asked to locate 12 vocabulary items within a play context of a
birthday party. Ten children were randomly assigned to each of 3 system
approaches: vocabulary in a grid format organized taxonomically, vocabulary in
a grid format organized schematically, and vocabulary in an integrated scene
organized schematically. The children participated in 4 learning and testing
sessions and 1 generalization session. Results indicated that the children per-
formed poorly in all conditions but were able to locate more vocabulary items in
the schematic scene condition than the taxonomic grid or schematic grid condi-
tions. There was evidence that the children failed to generalize their knowledge of
the vocabulary to facilitate learning of novel vocabulary items. The current design
of AAC dynamic display systems appears to be inappropriate for very young
children. Rather than relying solely on technology for these young children, early
intervention should target multiple modes of communication. AAC technologies
should be redesigned to reduce learning demands. Results are discussed with
implications for practice and suggestions for future research.
KEY WORDS: augmentative and alternative communication (AAC), assistive
technology, dynamic displays, children, learning
A
ugmentative and alternative communication (AAC) techniques
provide a means for young children to participate in social inter-
action and may facilitate children’s development of language, lit-
eracy, social, and cognitive skills. Computer-based AAC technologies offer
the advantages of programmable vocabulary storage and speech output,
allowing individuals increased potential power in communication. How-
ever, the current generation of AAC technologies is largely based on con-
ceptual models of adults who are not disabled; the representation, presen-
tation, organization, and selection of language concepts do not necessarily
Research• •Vol.
Vol.
4646• •298–312
298–312• •April
April 2003 • ©American Speech-Language-Hearing Association
2003
1092-4388/03/4602-0298
match the conceptual models of children (Light & Lind-
say, 1991). Because the instructional time and effort ex-
pended is in competition with other important opportuni-
ties for learning (e.g., language, literacy, and social skills;
Beukelman, 1991), examining alternative strategies for
symbol displays is an important topic. Ideally, AAC tech-
nologies would offer maximum power of communication
in exchange for minimal costs of learning; they would be
essentially transparent to the user (Light, 1997).
By the age of 30 months, typically developing chil-
dren are able to comprehend approximately 900 words
and produce approximately 500 words (Paul, 2001). They
can comprehend a variety of semantic/syntactic combina-
tions, including a variety of question forms (Miller & Paul,
1995; Paul, 2001). The ability to produce rhymes emerges
around this age, as well as many important morphologi-
cal markers, such as the present progressive (-ing), pos-
sessive (-’s), the contractible copula (he’s), and the use of
basic spatial terms (e.g., in, on, under) (Paul, 2001). Addi-
tionally, pragmatic skills are developing; children are able
to understand misrepresentations of reality, like lying or
teasing, and demonstrate increased use of language in
play at this age (Paul, 2001). Although most communica-
tion is grounded in the here-and-now context, talking
about objects or events absent from the immediate envi-
ronment emerges around 24 months and continues to in-
crease (Paul, 2001). Clearly, the language skills of 21/2-
year-old children, though still developing, represent a
complex and increasingly sophisticated system.
In order to meet these complex communication
needs, young children who require AAC need access to
a sufficient number of vocabulary concepts. Providing
access to a large number of items creates a significant
challenge for professionals. One way to meet this chal-
lenge is through dynamic display systems, in which a
selection from a display results in a new array of graphic
symbols (Lloyd, Fuller, & Arvidson, 1997). Electronic
dynamic display systems offer a number of options for
arranging symbols. One advantage to this type of dis-
play is that fewer symbols need to be on each page of
the display at one time, while still allowing access to a
larger vocabulary set. Having fewer symbols from which
to choose in a fixed display has resulted in increased
accuracy of symbol selection for young children (Mizuko,
Reichle, Ratcliff, & Esser, 1994). In a dynamic display
system, other symbols are not currently visible, but are
available through page linking. To use the system suc-
cessfully, however, it is helpful if the user understands
the overall organization of the system. An individual
using a dynamic display system must (a) understand
that other graphic symbols are available in the system,
even when they are not visible; (b) determine or remem-
ber the page where the item is located; and (c) deter-
mine or remember the correct selection to access the
desired message (Reichle, Dettling, Drager, & Leiter,
2000). These three conditions may be cognitively chal-
lenging for young children who require AAC. In particu-
lar, although the ability to talk about absent objects
emerges by the age of 24 months (Paul, 2001), most com-
munication by children this age is grounded in the im-
mediate environment. Understanding that symbols that
are not currently visible are still accessible may be very
difficult for these young children. The increasing use of
dynamic display systems has resulted in a need to in-
vestigate their efficacy with young children.
Approaches to AAC System Layout
To be accessed for communication, language con-
cepts must be placed within the physical “space” pro-
vided by an AAC system. Traditionally, AAC systems
have used a grid format, where graphic AAC symbols
representing individual concepts are located in separate
squares of a grid, organized into rows and columns. A
symbol selected from the grid in a dynamic display sys-
tem may result in another grid display with new vo-
cabulary. This type of layout may impose metalinguistic
demands beyond the skills of young children (Light &
Drager, in press). Literate individuals understand that
words are separate entities with individual meaning,
but young children may not be able to grasp the con-
cepts or “words” out of context until later stages of de-
velopment (Owens, 1996). There is no evidence that this
type of layout is appropriate for young children.
Recently, an integrated scenes (e.g., Companion soft-
ware1) option for AAC systems layout has been introduced.
In this layout, language concepts are embedded into con-
textual scenes. For example, a digital photograph may be
taken of a child’s room and stored in the system. Objects
and events within the photograph are then used as sym-
bols for communication. The concept play may be accessed
by selecting the toy box, whereas selecting an individual
toy may result in the name of the toy (e.g., blocks). This
type of layout may be used in a dynamic display system,
where the user would select links to successively narrow
the display to the target word (e.g., a house provides a
choice of rooms in a house, the kitchen leads to a repre-
sentation of a kitchen, and a milk bottle leads to the speech
output milk). Language concepts are represented within
a context, perhaps reducing metalinguistic demands and
facilitating early language learning. However, as the con-
cepts are embedded within specific contexts, it may be
more difficult for children to generalize vocabulary to new
contexts (Light & Drager, in press). For example, there
may not be a toy box at school, so “play” must be repre-
sented differently in that setting.
1
Companion software is manufactured by Assistive Technology, Inc.
Address: 7 Wells Avenue, Newton, MA 02459. Phone: 1-800-793-9227.
Web site: http://www.assistivetech.com.
Drager et al.: Performance of Young Children With AAC Systems 299
 Representations of concepts will necessarily differ
between these two types of layouts. Although concepts
represented in a grid layout are separated in physical
space, the symbols in an integrated scene derive their
meaning not only from their specific representation (e.g.,
a glass) but also from their relation within the rest of
the scene (e.g., a drink on a table surrounded by break-
fast foods, indicating orange juice). Results of research
by Lund, Millar, Herman, Hinds, and Light (1988) sug-
gest that children use a similar approach to represent-
ing concepts graphically. When asked to draw early
emerging concepts, children rarely drew isolated objects;
instead, they integrated the objects into scenes accom-
panied by rich explanations of occurring events. How-
ever, it is difficult to know how to embed abstract con-
cepts (such as more or what?) into integrated scenes,
much as it is difficult to represent these concepts graphi-
cally on grid-based systems.
Approaches to Language Organization
Within an AAC System Layout
Children may represent and link internal language
concepts in countless ways. However, these concepts must
be retrieved using AAC through specific, preprogrammed
selections. The organization of concepts in AAC is often
based on the way in which the adult designing the sys-
tem (clinician, parent, teacher, etc.) thinks about language.
In dynamic display systems, this organization is across
several “pages” of vocabulary. For a dynamic display sys-
tem to be used effectively, facilitators need to determine
an appropriate overall organization of the system. Light
(1992) has suggested that there are at least four primary
approaches to organizing vocabulary: schematic, taxo-
nomic, semantic–syntactic, and alphabetic.
A schematic approach to organization is based on
event schema (e.g., going to a party, ordering food, getting
ready for school). A variety of words associated with an
event, such as people, places, actions, and descriptors,
would be available on a given page, organized schemati-
cally. Typically developing children age 5 and under tend
to use a schematic approach to organization (Lucariello,
Kyratzis, & Nelson, 1992; Maaka & Wong, 1994; Nelson,
1996; Sell, 1992). As children get older, enter school, and
are exposed to more adult models, they learn to use a taxo-
nomic approach to organization, particularly for objects
that are highly familiar (Markman, Cox, & Machida,
1981). A taxonomic approach is based on hierarchical cat-
egories of words (e.g., people, places, toys). Although
younger typically developing children (under age 5) do
not tend to use a taxonomic approach to organizing vo-
cabulary, they are able to understand taxonomic catego-
ries, particularly when the objects are highly familiar and
the categories are labeled for them (Markman et al., 1981;
Lucariello et al., 1992; Krackow & Gordon, 1998).
300 Journal of Speech, Language, and Hearing Research • Vol. 46 • 298–312 • April 2003
A semantic–syntactic approach to organization
groups vocabulary according to parts of speech and their
relation within a syntactic framework. One example of
this approach is a Fitzgerald Key, where vocabulary
items are arranged left to right in the following order:
agent, action, descriptor, locative, and object (Bishop,
Rankin, & Mirenda, 1994). Often, this approach is used
to promote language development for children who are
encouraged to combine several words into sentences.
Finally, an alphabetic arrangement organizes vo-
cabulary alphabetically, which is not appropriate for
preliterate children.
Although research on typically developing children
suggests a developmental progression to concept organi-
zation, there is limited research regarding how language
should be organized on AAC systems. Additionally, most
of the research is based on how children organize nouns,
leaving much speculation as to how children would orga-
nize other concepts, such as relational terms (e.g., more),
descriptors (e.g., little), or question words (e.g., why?).
Fallon, Light, and Achenbach (2000) reported on research
investigating how typically developing 4- and 5-year-olds
organized line drawings that represented a variety of lan-
guage concepts (i.e., nouns, verbs, descriptors, questions).
The children had greater difficulty organizing abstract
concepts than concrete concepts. Their organizations
tended to be in two-word associations or small groups (3–
4 words) rather than larger groups (such as pages). The
majority of the organizations (over 80%) reflected sche-
matic organizations. For example, a child might group
“boy” and “pizza” and explain the combination by saying,
“This boy’s hungry so he’s going to eat some pizza.”
There has been no research to date investigating
the effects of different AAC system layouts and organi-
zations with young children. These aspects (layout and
organization of vocabulary within a layout) are expected
to affect the ease of learning systems and generaliza-
tion of this learning to new vocabulary. The current re-
search investigated the following questions:
1. How accurately can typically developing 21/2-year-
old children locate vocabulary prestored in AAC
technologies in the following conditions: vocabulary
in a grid format organized taxonomically, vocabu-
lary in a grid format organized schematically, and
vocabulary in an integrated scene organized sche-
matically?
2. Do the children’s performances improve across
learning sessions with these different organizations?
3. Do the children perform more accurately with con-
crete or abstract vocabulary?
4. Are the children able to generalize learning of the
organizations to facilitate learning of novel vocabu-
lary items?

Method
Participants
Thirty typically developing children were recruited
from day-care centers to participate. The children were
between the ages of 2 years 5 months and 2 years 11
months (M age = 2 years 8 months). Participants were
primarily from middle class families; approximately 10–
15% were from diverse ethnic and cultural backgrounds.
All participants had no prior experience with AAC tech-
nologies; had no identified speech, language, cognitive,
or physical disabilities; and had parental consent to
participate. The child’s ability to participate in the task
served as an informal screening for auditory compre-
hension. For all participants, the teacher and school
reports identified the children as within normal limits
in vision and auditory acuity (or vision corrected to
within normal limits). Ten children were randomly as-
signed to each of three conditions, with the constraints
that each group consisted of 5 boys and 5 girls and that
the mean ages of each group were equal.
Typically developing children were used in this in-
vestigation for several reasons. First, there is limited
empirical evidence to support using dynamic display
devices with young children. The current investigation
represented an initial exploration of how young children
process information in AAC systems from a developmen-
tal perspective. Using typically developing children to
determine the effects of the AAC organizations on learn-
ing without the confounding variables of motor, sensory
perceptual, and other impairments allowed for the es-
tablishment of a control group for a basis of comparison
for children with disabilities. It also allowed for a group
design, permitting the investigation of multiple factors
that will influence learning and gaining a developmen-
tal perspective on how young children learn these types
of systems.
Materials
Props
During the sessions, the children were introduced
to a birthday party context. This context was chosen
because it was familiar to the children, gender neutral,
and motivating. To set the context, a number of props
were available, including teddy bears (“Bobby the birth-
day boy” and his friends), party hats, cups, plates,
spoons, a wrapped present, and a variety of toys, such
as jacks, a toy car, blocks, and a book.
Vocabulary
Sixty-one vocabulary items were represented on
each system. Vocabulary items were selected from a com-
posite list of initial vocabulary words for preschoolers
(Beukelman & Mirenda, 1998) and supplemented with
vocabulary specific to a birthday party context. One vo-
cabulary item (blue) was used as an example. Of the
remaining 60 items, 30 vocabulary items were concrete
(i.e., concepts that involved a concrete referent or ac-
tion) and 30 were abstract (i.e., concepts that did not
involve a concrete object or action and that involved
abstract concepts such as relational terms, emotions,
and social greetings). These 60 represented a core num-
ber of items needed to assess performance across four
pages. Fewer than this number would result in very few
items on each “page.” All items were available on each
system at all times.
Twenty-four of the 60 items were targeted for in-
struction and generalization with the 21/2-year-old chil-
dren. Twelve of the 24 items were randomly assigned to
be vocabulary taught during learning sessions, with the
constraint that 6 were concrete and 6 were abstract. The
remaining 12 vocabulary items (6 concrete and 6 ab-
stract) were chosen as probe items during the generali-
zation session. Items were chosen that were develop-
mentally appropriate and that represented a variety of
vocabulary types (people, pronouns, prepositions, social
vocabulary, question words, verbs, and nouns). An addi-
tional constraint was that the number of words of each
vocabulary type was approximately equal between the
learning vocabulary and the generalization vocabulary.
The lists of concrete and abstract vocabulary words used
for learning appear in Appendix A. The lists of concrete
and abstract vocabulary used for generalization probes
appear in Appendix B.
The list of vocabulary was compared with norms for
the MacArthur Communication Development Invento-
ries (MCDI; Fenson et al., 1993), a commonly used in-
ventory assessing expressive and receptive vocabulary
for young children. Eleven of the 12 learning vocabu-
lary items (with the exception of win) and 11 of the 12
generalization vocabulary items (with the exception of
candles) are assessed in the MCDI. Norms indicate that
the majority of 30-month-old children (over 95%, on av-
erage) comprehend these items (Dale & Fenson, 1996).
AAC System Organizations
Three approaches to AAC system layout and vocabu-
lary organization were investigated. All approaches were
presented using a dynamic display consisting of a main
page with four links to four vocabulary pages, each with
12–20 symbols.
In the taxonomic grid organization, vocabulary
items were presented in a grid layout and organized
according to hierarchical categories (e.g., people, actions,
objects). Vocabulary items were accessed through a main
page with four symbols, each linked to a page of vocabu-
lary. The four page names were People, Things, Doing,
Drager et al.: Performance of Young Children With AAC Systems 301
and Sayings. Each page contained a grid of 5 × 5 boxes;
approximately 15 vocabulary items were represented in
the boxes on each page.
The schematic grid organization presented vocabu-
lary in a grid layout organized according to event
schemes. As with the taxonomic grid organization, a
main index page with four symbols was used to link to
each page of vocabulary items. The four page names were
Going to the Party, Eating Cake, Opening Presents, and
Playing Games. Each page also contained a grid of 5 × 5
boxes, with approximately 15 vocabulary items repre-
sented on each page. (See Appendix A for the locations
of individual vocabulary items.)
Both the taxonomic grid and the schematic grid or-
ganizations were presented on DynaVox systems
(DynaVox 2c, DynaVox 3100, or DynaVox software2).
Vocabulary items were represented with DynaSyms,3 as
this was the symbol set that was inherent in the sys-
tem. Appendixes A and B list the symbols that were used
for each vocabulary word.
The schematic scene organization presented vocabu-
lary within the contextual scenes of four rooms of a house.
Each of four rooms of a house linked to a representation
of that room with a colored integrated scene represent-
ing an event within a birthday party. The vocabulary
was represented graphically within each scene. The four
page names were Living Room (where the kids were ar-
riving at the party), Kitchen (where the kids were eat-
ing cake), Family Room (where the kids were opening
presents), and Playroom (where the kids were playing
games). On each page, vocabulary words were integrated
into the scene. See Figure 1 for an example of the vo-
cabulary page for the Family Room (although it should
be noted that the scene on the AAC device was in color).
The schematic scene organization was presented by Com-
panion software on a Freestyle computer.4 Scenes and
symbols representing vocabulary were drawn by an un-
dergraduate graphic design student and were scanned,
and incorporated into the software. In contrast to the
DynaVox systems, Companion software is not packaged
with a standard set of symbols. Appendixes A and B de-
scribe the line drawings that were used to represent each
vocabulary word within the integrated scenes.
All vocabulary, when selected, resulted in speech out-
put of the vocabulary item name. The DynaVox 2c and
3100 used the DECTalk, Kit the Kid voice. The Freestyle
2
DynaVox communication systems are manufactured by DynaVox, Inc., a
division of Sunrise Medical. Address: 2100 Wharton Street, Pittsburgh,
PA 15203. Phone: 1-888-697-7322. Web site: http://www.dynavoxsys.com.
3
DynaSyms are manufactured by DynaVox, Inc.
4
Freestyle systems are manufactured by Assistive Technology, Inc.
Address: 7 Wells Avenue, Newton, MA 02459. Phone: 1-800-793-9227.
Web site: http://www.assistivetech.com.
302 Journal of Speech, Language, and Hearing Research • Vol. 46 • 298–312 • April 2003
computer used the MacinTalk, Junior voice. The investi-
gator repeated all speech output from the devices to ens-
ure that the children understood the output.
Procedures
Each child participated in five individual sessions.
Sessions took place in the day-care setting or at the child’s
home. Whenever possible, sessions were conducted in
rooms separate from the other children in the center. Sev-
eral research assistants served as the interventionists;
typically, one interventionist was responsible for all data
collection sessions for each child. At the beginning of the
first session, the children were introduced to the teddy
bear (“Bobby”) and told that he couldn’t speak but had
the computer speak for him. They were asked to help
Bobby use his computer to speak. Then they were in-
structed on how to use the computer to locate a sample
vocabulary item, blue. Each child was asked if he/she could
see the pictures on the device and could hear the speech
output. The children were asked to pretend that it was
Bobby’s birthday and he was having a party.
Learning Sessions
The first three learning sessions were 2–4 days
apart. A fourth session was scheduled 2 weeks after the
third learning session, to determine whether the chil-
dren maintained performance. In each of these four ses-
sions, children were asked to locate the 12 vocabulary
items within a play context of the birthday party. The
order of target items was randomly selected for each
session from five randomly determined orders. The in-
structional prompt for the target item was given by set-
ting a context for the children (e.g., “Bobby wants to
talk to his Mommy. Please show him how to say
MOMMY.”). The last word in the instructional prompt
served as the cue for the child to find the word (e.g.,
MOMMY). The interventionist secured the child’s at-
tention for each target word, and the instructional
prompts were stated once and were not repeated. The
instructional prompts were the same for all three ex-
perimental conditions (see Appendix C for examples).
The props were used to set the context, act out vo-
cabulary, and provide a period of play after correct re-
sponses. The children were given feedback for incorrect
or no responses. During this feedback, the instructors
showed the children the location of the target item and
explained (a) the page link location, (b) the rationale for
the page representation, (c) the item location, and (d)
the rationale for the item representation. Appendix C
shows some examples of feedback for vocabulary in each
of the experimental conditions. After the instructor pro-
vided this feedback, he or she used the props to act out
vocabulary. All 12 vocabulary items were probed within
Figure 1. Example of a vocabulary page from the schematic scene display. The scene was in color on the system.
each session. Each session was approximately 20–30
min. All sessions were audiotaped.
Generalization Session
The generalization session was organized to assess
the children’s generalization to new vocabulary items and
generalization of the vocabulary (taught in the learning
sessions) to functional use in play situations. To assess
whether the children could generalize knowledge of spe-
cific vocabulary to the organization of the system, novel
vocabulary items were probed during one session that took
place 2–4 days after the third learning session. These vo-
cabulary items had always been available in the system
but were never probed. Procedures were identical to the
learning sessions, except that during feedback for incor-
rect or no responses, the children were shown the loca-
tion of the item but not provided with the rationale for
the location and representation. This part of the generali-
zation session took approximately 20 min.
To assess whether the children were able to gen-
eralize their knowledge of the vocabulary items to
functional use within a free play situation, each child
was engaged in a 6-min free play activity during the
second half of the generalization session. The instruc-
tor told each child, “Now we can play at Bobby’s birth-
day party. Remember to help Bobby talk by pushing
the buttons on the computer.” Procedures recom-
mended by Miller (1981) for eliciting spontaneous
language samples were used. Within the free play,
the instructor suggested three play situations relat-
ing to the birthday party context (i.e., playing games,
eating cake and ice cream, and opening presents). At
approximately 1-min intervals, the instructor
prompted the child to use the AAC system to help
Bobby talk (i.e., “What would Bobby say?”). Each se-
lection on the AAC system was recorded as sponta-
neous or following a prompt.
Procedural Reliability
Before data collection, each instructor was trained
on the procedures until at least 90% of the steps were
completed correctly (over 10 items). To maintain proce-
dural integrity, an independent judge determined that
Drager et al.: Performance of Young Children With AAC Systems 303
the instructor followed the appropriate procedural se-
quence for each of five vocabulary items checked at ran-
dom for each session. Using the formula (number of steps
followed correctly / total number of steps) × 100, proce-
dural reliability was 98.4% (range = 80–100%). An im-
mediate review of procedures was undertaken if proce-
dural accuracy was less than 90% for any instructor.
Response Scoring
The dependent variable was the children’s accuracy
in locating target vocabulary items. A response was
scored as correct if the speech output of the device
matched the target vocabulary item. A response was
scored as incorrect if the speech output of the device did
not match the target item. If the child made an acciden-
tal selection and attempted to self-correct, the first re-
sponse that resulted in speech output was still used to
determine correctness. A response was scored as a “no
response” if the child did not make a selection resulting
in speech output within 60 sec.
Reliability
To evaluate the reliability with which the instructor
recorded the accuracy of the children’s responses, an in-
dependent judge listened to five randomly selected items
per session (42%) on the audiotapes and scored the re-
sponses as correct, incorrect, or no response. Reliability
was evaluated on 143 sessions (of 150 total); during 7 ses-
sions problems with the audiotape players prohibited re-
cording for reliability purposes. Using the formula (agree-
ments / agreements + disagreements) × 100, the
point-to-point reliability was 99.4%, indicating that the
scoring of the dependent measure was highly reliable.
Results
Learning
Figure 2 shows the learning curves for the three
groups of participants across sessions. During the first
session, the children were able to locate a mean of 0.9
items out of 12 (0.8 for the taxonomic grid system, 0.7 for
the schematic grid system, and 1.3 for the schematic scene
system). The children had a mean of less than 8% correct
for Session 1. Although there was some evidence of learn-
ing, all children demonstrated very low levels of accuracy
with all systems across the four sessions. After four ses-
sions (approximately 2 hr) of individual instruction, par-
ticipants in the taxonomic grid condition correctly selected
an average of 1.6 items out of 12 (SD = 1.96; range = 0 to
4), participants in the schematic grid condition correctly
selected an average of 1.9 items (SD = 1.64; range = 0
to 5), and participants in the schematic scene condition
correctly selected an average of 4.1 items (SD = 2.85;
range = 0 to 8).
The main effect for system organization was statisti-
cally significant, F(2, 27) = 5.508, p = .01. Planned post
hoc analyses of group differences were accomplished with
three independent-samples t tests. The following results
were statistically significant: participants using the sche-
matic scene system performed better than participants
using the taxonomic grid condition, t(18) = –3.326, p =
.004, and better than participants using the schematic
grid condition, t(18) = –2.138, p = .046. If alpha error rate
for individual tests is adjusted (.05/3 = .017) to create a
more conservative family-wise error rate of .05, the dif-
ference between the schematic scene condition and the
schematic grid condition was not statistically significant.
The main effect for session was statistically signifi-
cant, F(3, 81) = 8.649, p < .0001. On average, participants

Data Analysis
Figure 2. Performance of 21/2-year-old children during learning
Two separate analyses of variance (ANOVAs) were sessions for the taxonomic grid, schematic grid, and schematic
used to evaluate the effects of system organization, ses- scene system organizations.
sions, and vocabulary type. The first ANOVA used a three
factor mixed design to evaluate the children’s learning of
the vocabulary items. The between-subjects factor was
system organization (taxonomic grid, schematic grid, and
schematic scene). The within-subjects factors were ses-
sion (Session 1, Session 2, Session 3, and maintenance)
and vocabulary type (concrete, abstract). A second ANOVA
used a two factor mixed design to examine whether the
children were able to generalize their learning to novel
vocabulary. The between-subjects factor was system or-
ganization (taxonomic grid, schematic grid, and schematic
scene). The within-subjects factor was session (Learning
Session 1, generalization session).

304 Journal of Speech, Language, and Hearing Research • Vol. 46 • 298–312 • April 2003
performed better across successive sessions. The mean
gain in vocabulary items from the first to the last session
for the taxonomic grid condition was 0.9 items (SD = 2.08;
range = –1 to +5), the mean gain for the schematic grid
condition was 1.2 items (SD = 1.69; range = –1 to +4), and
the mean gain for the schematic scene condition was 2.8
items (SD = 3.01; range = –4 to +7). However, a total of 5
children (17%) did not learn any items across the four
sessions (0% accuracy across all sessions): 2 taxonomic
grid participants, 2 schematic grid participants, and 1
schematic scene participant. Fourteen participants (47%)
did not show any improvement in performance at the last
session (maintenance) from the first session.
The main effect for vocabulary type was statistically
significant, F(1, 27) = 61.868, p < .0001. Participants
learned more concrete vocabulary (M = 1.5 items, SD =
1.54) than abstract vocabulary (M = 0.2 items,
SD = 0.58) across all systems and sessions. None of the
children successfully located any abstract items during
the first session, and over half of the children (60%) did
not learn any abstract vocabulary items during any of
the learning sessions. There were no significant inter-
actions between the factors.
Generalization
Generalization to New Vocabulary
Figure 3 shows the performance of each group for
the first learning session (Session 1) and the Generaliza-
tion Session. Participants showed no evidence of gener-
alization of system organization to novel vocabulary. The
main effects for system organization and for session were
not statistically significant. There was a statistically sig-
nificant Session × System Organization interaction,
Figure 3. Performance of 21/2-year-old children during the first
learning session and the generalization session for the taxonomic
grid, schematic grid, and schematic scene system organizations.
F(2, 27) = 4.359, p = .023. The mean difference between
the first session and the generalization session for the
taxonomic grid condition was 0.4 items out of 12 (SD =
0.84; range = –1 to +2). The mean difference for the sche-
matic grid condition was 0.2 items out of 12 (SD = 1.14;
range = –1 to +2). For both of these organizational strat-
egies, participants did as well or slightly better with the
generalization vocabulary than with the initial vocabu-
lary. In contrast, the mean difference for the schematic
scene condition was –0.8 items out of 12 (SD = 0.92;
range = –2 to 0). Participants did better during the first
session than during generalization. It should be noted
that the mean differences across all conditions were ap-
proximately only one item.
Functional Use in Free Play
During the free play context, the children used the
AAC systems to select an average of 0.3 words spontane-
ously. During the 6 min of play, children spontaneously
selected an average of 0.2 words (range = 0–1) in the taxo-
nomic grid condition and 0.7 words (range = 0–6) in the
schematic grid condition. No children in the schematic
scene condition spontaneously selected vocabulary dur-
ing the free play. Rates of system use were low even af-
ter the instructor prompted the children to use the sys-
tem. An average of 0.5 words were selected: 0.8 (range =
0–4) in the taxonomic grid condition, 0.3 (range = 0–1)
in the schematic grid condition, and 0.4 (range = 0–2) in
the schematic scene condition. Even with prompting by
the instructor, 20 children (67%) did not use the system
at all during the free play.
Discussion
Accuracy of 21/2-Year-Olds in Learning
Different AAC System Organizations
The most salient finding of the current investigation
was that across all the organizations, 21/2-year-old chil-
dren had great difficulty learning to locate target vocabu-
lary on dynamic display systems. Successful use of AAC
dynamic display systems requires extensive time and prac-
tice by children—time that may be better spent learning
language, literacy, socialization, and play skills.
It is possible that the systems were, in fact, appropri-
ate for the young children, whereas the assessment task
was inappropriate. To assess this hypothesis, a pilot study
was conducted to determine if the children could perform
the task without the AAC device. Each child was provided
with the eliciting context for the 12 learning vocabulary
items, either at the end of the final session or during an
additional session. The procedure was identical to the first
three sessions except that the child was allowed to re-
spond orally, and no feedback was given for incorrect re-
Drager et al.: Performance of Young Children With AAC Systems 305
sponses. All children were able to perform the task (i.e.,
able to tell Bobby how to say the target word) with 100%
accuracy orally. These results suggest that the children
were able to handle the process without the AAC system.
The addition of AAC system use presented considerable
demands on the user, particularly for the novice user.
As the present study did not consider the use of static
displays, it is impossible to determine how the children
would have performed. There is evidence that the demands
of a dynamic display system presented some difficulty for
the children, either in locating the correct page or in rec-
ognizing the representations used as links to the correct
page. In the present study, the probability of choosing the
correct page due to chance responding was .25 (of four
page links). In a post hoc error analysis, children in all
three conditions had a probability of .22–.28 of choosing
the correct page during the first session. The system or-
ganization was not transparent to the children, even af-
ter four sessions. The children in the schematic grid and
taxonomic grid conditions had a .23 and .32 probability,
respectively, of choosing the correct page, which is still at
approximately chance level. The children in the schematic
scene condition had a probability of .52 of choosing the
correct page during the fourth learning session, suggest-
ing that these children were beginning to learn the pages
on which items were located.
Once the children were on the correct page, the prob-
ability of choosing the correct item by chance was ap-
proximately .07. For the children in this study, this step
of using a dynamic display appeared to be easier than
choosing the correct page. For all three conditions, the
children had a probability of greater than .40 of choos-
ing the correct item, once they were on the correct page.
However, these results should be interpreted with cau-
tion given the small number of instances in which the
children got to the correct page.
Differences Among System Organizations
Effect of System Layout
The children in the schematic scene group did sig-
nificantly better than the children in the two groups
with the grid layout. There are several possible expla-
nations for this result. One possibility is that embed-
ding language concepts within contextual scenes re-
duced the metalinguistic demands of the task for the
children. Words were not presented as isolated concepts,
as they were in the two grid conditions, but presented
within contexts. This may be more similar to the con-
ceptual maps of children.
Another possible explanation for the advantage of the
schematic scene condition relates to the issue of working
memory in children. The task of finding an item in a grid
system layout places a large demand on working memory.
306 Journal of Speech, Language, and Hearing Research • Vol. 46 • 298–312 • April 2003
In the schematic scene condition, the children may have
been able to chunk together vocabulary items that were
represented together within a scene, decreasing the work-
ing memory demands for those items.
Another explanation for the better performance in the
schematic scene condition is the representation used for
the page links from the main page. For the taxonomic
grid and schematic grid conditions, the main pages con-
sisted of four link buttons with single icons representing
each page of vocabulary. For example, in the taxonomic
grid condition, the icon representation for the People page
was a picture of a group of people. In the schematic scene
condition, the representation on the main page was the
context of rooms of the house where the birthday party
was being held. To access each “room”, the child selected
a representation of the room from the main page: a graphic
of a house. These room representations were smaller ver-
sions of the actual page of vocabulary, making the process
of finding the correct page much like a matching-to-sample
task. Some vocabulary items were recognizable in these
smaller versions. Although the most common error for all
the children was to select the wrong page when they were
incorrect, the children in the schematic scene group more
often chose the correct page than the children in either
the schematic grid or taxonomic grid groups. These chil-
dren had a better than chance probability of choosing the
correct page (.52, with chance at .25). Once they were on
the correct page, they had no higher a probability of choos-
ing the correct item than the children in the other two
groups. Thus, the main difference between the schematic
scene condition and the two grid conditions was the prob-
ability of getting to the correct page.
The advantage seen for the schematic scene condi-
tion for typically developing children may be mitigated
by other factors present for children with disabilities.
Contextual scenes may be more visually complex than
single icons, increasing the difficulty level for children
with visual–spatial disabilities. Also, it is possible that
this approach places a higher motoric burden on the AAC
user because the concepts are no longer discrete from
one another. Additionally, for children not able to use
direct selection, it may be more difficult to visually fol-
low a scanning pattern throughout the selections when
items are not located in a grid format.
Effect of System Organization
For the two grid conditions, there was no advantage
to a schematic organization over a taxonomic organiza-
tion. Although this result may be surprising, as a devel-
opmental perspective suggests that 21/2-year-olds would
do better with a schematic organization, it is consistent
with the literature stating that young children are able to
understand taxonomic categories when the objects are
familiar and the categories are labeled for them (Krackow
& Gordon, 1998; Lucariello et al., 1992; Markman et al.,
1981). There are several possible explanations for the lack
of an advantage for the schematic organization in a grid
format for the present study. First, the schema were de-
veloped by the researchers (i.e., adults) and may not re-
flect typical organizations of young children. Second,
schema were the same for each child in the relevant group;
they were not personalized for each child.
Generalization of Learning to New
Vocabulary
On average, the children did not show any evidence
of generalization to novel vocabulary on any type of dis-
play or layout. This result is not surprising, as the chil-
dren learned very few items. As with the errors during
learning sessions, a post hoc error analysis showed that
the children were more likely to select an item on an
incorrect page (a page different than the page of the
target item) than to make an incorrect selection on the
correct page.
Clinical and Educational Implications
Although there is some evidence that a schematic
scene approach may be easier for some young children to
learn, it is clear that the current design of AAC technolo-
gies is not transparent to 21/2-year-old children without
more intensive intervention than was provided in the cur-
rent investigation. While children are learning these sys-
tems, it is critical that they be provided with effective
modes of communication. Exploring communication
means such as vocalizations, gestures, and low technol-
ogy systems (e.g., boards or books) should be a priority for
any young child who uses AAC. Concurrently, instruction
can be carried out on appropriate AAC technologies.
One possible solution to the problem of the signifi-
cant learning demands of AAC technologies for young
children is to redesign AAC systems to be more appro-
priate for the conceptual models of young children. An-
other solution is to change the instruction and/or the
support provided to the child during the learning and
use of AAC technologies. Given the systems’ significant
lack of transparency, caregivers and educators can pro-
vide external scaffolding support to reduce the learning
demands on the child. For example, rather than placing
the burden of determining the organization on the child,
the adult may manage the “paging” system for the child
during instruction.
Limitations of Current Research
Although this study provides us with an initial un-
derstanding of the learning demands of AAC technolo-
gies for 21/2-year-olds, there are some limitations to the
research that must be considered. First, typically de-
veloping children participated rather than children with
communication disabilities. The results provide impor-
tant insight into developmental issues; however, the
outcomes may not be fully generalizable to children with
disabilities. Clearly, subsequent work investigating the
learning and use of AAC systems with children with dis-
abilities is critical. Additionally, the children’s language
skills were not tested individually. Consequently, the
possibility exists that some of the children did not have
the language skills expected of typically developing chil-
dren of that age.
Another potential limitation to the present study is
that the conditions differed on more than one attribute.
The schematic grid and taxonomic grid systems used
the same symbol representations, whereas the schematic
scene system used different representations. These dif-
ferences were a result of the different approaches to lan-
guage organization taken by the systems. These differ-
ent representations may have had an effect on the
results, although the children in the schematic scene
condition did not have a higher probability of choosing
the correct representation once they were on the correct
page than the children in the other two conditions. Ad-
ditionally, the schema for the schematic grid condition
were based on events, while the schema for the sche-
matic scene condition were based on places where these
events took place. It is unclear whether this difference
affected the results. The hardware used for each condi-
tion also differed, although each system used a touch
screen for access. It is difficult to isolate the effects of
these differences.
Other limitations include the instructional tech-
nique, the number of sessions, and the type and num-
ber of vocabulary items. The present study used instruc-
tion within a functional play context. It is possible that
the children’s performance may have been increased
with other instructional techniques (e.g., child-centered
approach with modeling or direct instruction) or with
other instructional steps (e.g., teaching one symbol at a
time or one page at a time). For example, another strat-
egy may have been to first teach the children more in-
formation about how the vocabulary was organized in
general (e.g., the People page) and then teaching spe-
cific items within that page. This type of instruction may
have led to different results in generalization to new
vocabulary.
Another potential limitation is that the children
were involved in only four learning sessions. It is pos-
sible that learning curves would have varied after a
greater number of sessions or more extended practice
with the items, and that the critical amount of learning
or exposure had not yet taken place after four sessions.
Drager et al.: Performance of Young Children With AAC Systems 307
It would be interesting to determine if more sessions
were necessary or if performance would never reach a
reasonable level.
The abstract vocabulary items were difficult for the
children, and the use of abstract vocabulary significantly
affected the children’s performance. However, young
children begin to use and understand abstract vocabu-
lary at an early age (Fenson et al., 1993).
Finally, the present study taught 12 vocabulary
items within four pages, with a total of 60 symbols. It is
likely that children would have performed differently
had they had been instructed on fewer or more items.
Future Directions
Future research is required to improve AAC tech-
nologies for young children with significant communi-
cation disabilities. Investigations should be made to
delineate the reasons why the schematic scene condi-
tion was easier for some 21/2-year-olds to learn than the
two grid conditions. Research should investigate the
effectiveness of strategies that make the layouts and
organizations of AAC technologies more transparent to
young children (e.g., the main page of the schematic
scene condition contained icons that were smaller ver-
sions of the pages they were linked to). It would also be
interesting to see if different results would be seen if
the children had more extensive practice and instruc-
tion. Additionally, future research is required to inves-
tigate (a) young children’s representations and organi-
zations of picture-based vocabularies, (b) young
children’s performance with approaches to AAC orga-
nizations other than dynamic display (e.g., fixed dis-
play, or several fixed displays available at one time), (c)
the effectiveness of different instructional techniques
and procedures to teach AAC technologies to young chil-
dren, (d) new approaches to AAC design that better re-
flect the developmental organizations of young children,
and (e) learning demands of aspects of AAC technolo-
gies other than layout or organization (e.g., selection
techniques).
Additionally, children go through significant devel-
opmental changes before school age. Future research
should include older children to determine the develop-
mental trend for this type of task. It will be important
to establish the age at which typically developing chil-
dren can easily perform this task and use these types of
systems. Similar investigations by the authors are cur-
rently underway with 3- to 5-year-old children to deter-
mine the role of development in performance.
The present study investigated learning demands
on the production of vocabulary, as the majority of AAC
systems are used primarily for production. However,
these systems can be valuable tools to augment com-
308 Journal of Speech, Language, and Hearing Research • Vol. 46 • 298–312 • April 2003
prehension as well (e.g., Romski, Sevcik, & Adamson,
1997). Future research should include comprehension
measures in addition to production measures.
The idea of using technology to enhance the com-
munication of young children who require AAC is capti-
vating. However, these technologies need to offer maxi-
mum power (e.g., allow young children to fulfill
communication needs, facilitate receptive and expres-
sive language, and facilitate future literacy development)
with as minimal “costs of learning” as possible (Light,
1997). The present research has attempted to determine
some of the costs of learning dynamic display systems
for 21/2-year-old children. Future research is urgently
needed to further delineate these costs and to investi-
gate alternatives that will decrease these costs and in-
crease the magic of communication for young children
with significant communication disabilities.
Acknowledgments
This research was done as part of the Communication
Enhancement Rehabilitation Engineering Research Center
(AAC-RERC), which is funded by the National Institute on
Disability and Rehabilitation Research of the U.S. Depart-
ment of Education under grant number H133E980026. The
opinions contained in this publication are those of the
grantee and do not necessarily reflect those of the Depart-
ment of Education. Additional information on the AAC-
RERC is available at http://www.aac-rerc.org/.
This article is based in part on a presentation at the
Ninth International Conference of the International Society
for Augmentative and Alternative Communication, Washing-
ton, DC, August 2000.
References
Beukelman, D. (1991). Magic and cost of communicative
competence. Augmentative and Alternative Communica-
tion, 7, 2–10.
Beukelman, D., & Mirenda, P. (1998). Augmentative and
alternative communication: Management of severe
communication disorders in children and adults. Balti-
more: Paul H. Brookes.
Bishop, K., Rankin, J., & Mirenda, P. (1994). Impact of
graphic symbol use on reading acquisition. Augmentative
and Alternative Communication, 10, 113–125.
Dale, P. S., & Fenson, L. (1996). Lexical development
norms for young children. Behavioral Research Methods,
Instruments, and Computers, 28, 125–127.
Fallon, K., Light, J., & Achenbach, A. (2000, August). The
semantic organization patterns of young children: Implica-
tions for AAC. Paper presented at the biennial conference
of the International Society for Augmentative and
Alternative Communication, Washington, DC.
Fenson, L., Dale, P. S., Reznick, J. S., Thal, D., Bates,
E., Hartung, J. P., et al. (1993). The MacArthur Commu-
nicative Development Inventories: User’s Guide and
Technical Manual. San Diego, CA: Singular.
Krackow, E., & Gordon, P. (1998). Are lions and tigers
substitutes or associates? Evidence against slot filler
accounts of children’s early categorization. Child Develop-
ment, 69, 347–354.
Light, J. (1992, November). Cognitive science: Implications
for clinical practice in AAC. Miniseminar presented at the
annual convention of the American Speech-Language-
Hearing Association, San Antonio, TX.
Light, J. (1997). “Let’s go star fishing”: Reflections on the
contents of language learning for children who used
aided AAC. Augmentative and Alternative Communica-
tion, 13, 158–171.
Light, J. C., & Drager, K. D. R. (in press). Improving
the design of augmentative and alternative technologies
for young children. Assistive Technology.
Light, J., & Lindsay, P. (1991). Cognitive science and
augmentative and alternative communication. Augmenta-
tive and Alternative Communication, 7, 186–203.
Lloyd, L., Fuller, D., & Arvidson, H. (1997). Augmentative
and alternative communication: A handbook of principles
and practices. Boston: Allyn & Bacon.
Lucariello, J., Kyratzis, A., & Nelson, K. (1992). Taxo-
nomic knowledge: What kind and when. Child Develop-
ment, 63, 978–998.
Lund, S., Millar, D., Herman, M., Hinds, A., & Light, J.
(1998, November) Children’s pictorial representations of
early emerging concepts: Implications for AAC. Poster
presented at the annual convention of the American
Speech-Language-Hearing Association, San Antonio, TX.
Maaka, M., & Wong, E. (1994, April). Semantic memory in
young children: The script-based categorization of early
words. Paper presented at the annual meeting of the
American Educational Research Association, New
Orleans, LA.
Markman, E. M., Cox, B., & Machida, S. (1981). The
standard object-sorting task as a measure of conceptual
organization. Developmental Psychology, 17, 115–117.
Miller, J. M. (1981). Assessing language production in
children: Experimental procedures. Needham Heights,
MA: Allyn & Bacon.
Miller, J., & Paul, R. (1995). The clinical assessment of
language comprehension. Baltimore: Paul H. Brookes.
Mizuko, M., Reichle, J., Ratcliff, A., & Esser, J. (1994).
Effects of selection techniques and array sizes on short-
term visual memory. Augmentative and Alternative
Communication, 10, 237–244.
Nelson, K. (1996). Language in cognitive development:
Emergence of the mediated mind. New York: Cambridge
University Press.
Owens, R. (1996). Language development: An introduction
(4th ed.). Needham Heights, MA: Allyn & Bacon.
Paul, R. (2001). Language disorders: From infancy through
adolescence. Philadelphia: Mosby.
Reichle, J., Dettling, E. E., Drager, K. D. R., & Leiter, A.
(2000). A comparison of correct responses and response
latency for fixed and dynamic displays: Performance of a
learner with severe developmental disabilities. Augmenta-
tive and Alternative Communication, 16, 154–163.
Romski, M. A., Sevcik, R. A., & Adamson, L. B. (1997).
Framework for studying how children with developmen-
tal disabilities develop language through augmented
means. Augmentative and Alternative Communication,
13, 172–178.
Sell, M. (1992). The development of children’s knowledge
structures: Events, slots, and taxonomies. Journal of
Child Language, 19, 659–676.
Received May 5, 2002
Accepted October 8, 2002
DOI: 10.1044/1092-4388(2003/024)
Contact author: Kathryn D. R. Drager, PhD, Pennsylvania
State University, Department of Communication Sciences
and Disorders, 110 Moore Building, University Park, PA
16802. E-mail: kdd5@psu.edu
Drager et al.: Performance of Young Children With AAC Systems 309
Appendix A. Vocabulary during learning sessions, with location and representation of symbols, for
each AAC system organization.
Vocabulary Schematic scene (colored line drawings
item Taxonomic grid (DynaSym) Schematic grid (DynaSym) drawings as part of a larger scene)

Concrete vocabulary

Baby People page (BABY) Going to the Party page (BABY) Living Room page (crawling baby)
Ball Thing page (BALL) Playing Games page (BALL) Playroom page (ball)
Book Thing page (BOOK) Opening Presents page (BOOK) Living Room page (book)
Cake Thing page (CAKE) Eating Cake page (CAKE) Kitchen page (cake on table)
Dog People page (DOG) Going to the Party page (DOG) Living Room page (dog)
Mommy People page (MOMMY) Opening Presents page (MOMMY) Kitchen page (young woman)

Abstract Vocabulary

Come Action page (COME) Going to the Party page (COME) Living Room page
(girl motioning from door)
In Action page (IN) Going to the Party page (IN) Kitchen page
(girl pouring juice into glass)
On Action page (ON) Eating Cake page (ON) Family Room page
(cat sitting on shelf)
Please Saying page (PLEASE) Eating Cake page (PLEASE) Kitchen page
(boy with hands in prayer form)
What Saying page (WHAT) Opening Presents page (WHAT) Family Room page
(boy examining unopened present)
Win Action page (WINNER) Playing Games page (WINNER) Playroom page
(smiling girl with arms lifted,
holding a blue ribbon)

Appendix B (p. 1 of 2). Vocabulary during generalization session, with location and representation of symbols, for
each AAC system organization.

Vocabulary Schematic scene (colored line drawings

item Taxonomic grid (DynaSym) Schematic grid (DynaSym) drawings as part of a larger scene)

Concrete vocabulary
Candles Thing page (CANDLES) Eating Cake page (CANDLES) Kitchen page
(lit candles on cake)
Chair Thing page (CHAIR) Eating Cake page (CHAIR) Kitchen page (chair)
Daddy People page (DADDY) Opening Presents page (DADDY) Kitchen page (young man)
Friend People page (FRIEND) Playing Games page (FRIEND) Living Room page
(girl talking closely with another)
Home Thing page (HOUSE) Going to the Party page (HOUSE) Family Room page
(picture of house and tree on wall)
Ice Cream Thing page (ICE CREAM) Eating Cake page (ICE CREAM) Kitchen page
(sundae with cherry)
Abstract vocabulary
Hi Saying page (HI) Going to the Party page (HI) Living Room page
(smiling girl waving arms)
Look Action page (LOOK) Opening Presents page (LOOK) Family Room page
(boy watching another
opening present)

310 Journal of Speech, Language, and Hearing Research • Vol. 46 • 298–312 • April 2003
Appendix B (p. 2 of 2). Vocabulary during generalization session, with location and representation of symbols, for
each AAC system organization.

Vocabulary Schematic scene (colored line drawings

item Taxonomic grid (DynaSym) Schematic grid (DynaSym) drawings as part of a larger scene)

Me People page (ME) Playing Games page (ME) Living Room page
(boy pointing to himself)
Out Action page (OUT) Going to the Party page (OUT) Kitchen page
(window with view of outdoors)
Play Action page (PLAY) Playing Games page (PLAY) Playroom page
(girl with blindfold pinning
tail on donkey)
Where Saying page (WHERE) Opening Presents page (WHERE) Playroom page
(boy with confused face shrugging)

Appendix C (p. 1 of 2). Examples of the eliciting context for the target vocabulary and the feedback for the three
system layouts.

Target word Context Taxonomic grid Schematic grid Schematic scene

CAKE Bobby wants a piece of cake. We need to find CAKE.
Please show him how to A cake is a thing we eat.
say CAKE. All the things are on the
THINGS page because we
can put lots of things in
this box.
This is CAKE because it’s
a picture of a cake
that we can eat.
We need to find CAKE.
Cake is on the EATING
CAKE page because
everyone is eating
birthday cake.
This is the EATING CAKE
page because there are
people at the talbe ready
to eat cake.
This is CAKE because it’s
a picture of a cake
that we can eat.
We need to find CAKE.
Cake is in the KITCHEN
because that’s where
the kids are eating the
birthday cake.
This is the KITCHEN page
because it’s a picture of
the kids eating cake in
the kitchen.
This is CAKE because it’s a
birthday cake that we
can eat.

BABY Bobby’s baby sister is crying. We need to find BABY.
Please show him how to A baby is a person.
say BABY. All the people are on
the PEOPLE page.
This is the PEOPLE page
because this is a picture
of a lot of people.
This is BABY because it’s
a picture of a little baby
sleeping.
We need to find BABY.
Baby is on the GOING
TO THE PARTY page
because the baby is
going to the party.
This is the GOING TO THE
PARTY page because
this is a picture of a
person going into the
house for the party.
This is BABY because it’s
a picture of a baby
sleeping.
We need to find BABY.
Baby is in the LIVING
ROOM because the baby
is excited about the kids
coming to the party.
This is the LIVING ROOM
page because it’s a picture
of the kids coming into the
living room for the party.
This is BABY because it’s a
baby crawling on the
floor.

Drager et al.: Performance of Young Children With AAC Systems 311

Appendix C (p. 2 of 2). Examples of the eliciting context for the target vocabulary and the feedback for
the three system layouts.

Target word Context Taxonomic grid Schematic grid Schematic scene

BALL Bobby loves to play ball. We need to find BALL.
Please show him how to A ball is a thing we play with.
say BALL. All the things are on the
THINGS page.
This is the THINGS page
because we can put lots
of things in this box.
This is BALL because it’s a
picture of a ball that we can
bounce.
We need to find BALL.
Ball is on the PLAYING
GAMES page because
some games are played
with a BALL.
This is the PLAYING
GAMES page because
it’s a picture of a spinner
from a game.
This is BALL because it’s
a picture of a ball that
we can bounce.
We need to find BALL.
Ball is in the PLAYROOM
because we might play ball
in the playroom.
This is the PLAYROOM
page because it’s a
picture of the kids
playing in the playroom.
This is BALL because it’s
a ball we can bounce.

MORE Bobby wants to blow We need to find MORE.
more bubbles. More is something we say
Please show him how when we want more.
to say MORE. More is something we say,
so it’s on the SAYING page.
This is the SAYING page
because it’s a picture of
people saying things.
This is MORE because there
are more marbles on this
side (point).
We need to find MORE.
More is on the EATING
CAKE page because the
kids want more cake.
This is the EATING
CAKE page because
there are people at the
table ready to eat cake.
This is MORE because
there are more marbles
on this side (point).
We need to find MORE.
More is in the FAMILY
ROOM because Bobby
wants to open more
presents.
This is the FAMILY ROOM
page because it’s a picture
of the kids opening presents
in the family room.
This is MORE because there
are more presents here.

IN The kids are playing outside.
It’s time to go in the house.
Please show Bobby how
to say IN.
We need to find IN.
In is a place where we can
put something.
Putting something in is
something we do.
All the things we do are
on the DOING PAGE.
This is the DOING PAGE
because it’s a picture of
a person doing lots
of things.
This is IN because it’s a picture
picture of a ball in a box.
We need to find IN.
In is on the GOING
TO THE PARTY page
because everybody will
come in the house
for the party.
This is the GOING TO
THE PARTY page
because this is a picture
of a person going into
the house for the party.
This is IN because it’s a
picture of a ball in a box.
We need to find IN.
In is in the KITCHEN
page because someone
is pouring juice in a cup.
This is the KITCHEN page
because it’s a picture
of the kids eating cake
in the kitchen.
This is IN because
a girl is pouring
juice in the cup.

WHAT Bobby can’t hear Suzy talking. We need to find WHAT.
Please show him how to What is something we say
say WHAT. to ask a question.
What is something we say
so it’s on the SAYING page.
This is the SAYING page
because it’s a picture of
people saying things.
This is WHAT because it’s
a picture of a box with a
question mark on it.
We don’t know what’s
in the box.
We need to find WHAT.
What is on the OPENING
PRESENTS page because
Bobby wonders what is
inside each present.
This is the OPENING
PRESENTS page because
it’s a picture of a ribbon
to wrap presents.
This is WHAT because
it’s a picture of a box
with a question mark
on it.
We don’t know what’s
in the box.
We need to find WHAT.
What is in the FAMILY
ROOM because the kids
are asking Bobby,
“What did you get?”
This is the FAMILY ROOM
page because it’s a
picture of the kids
opening presents in the
family room.
This is WHAT because
this boy is asking,
“What is in the box?”

312 Journal of Speech, Language, and Hearing Research • Vol. 46 • 298–312 • April 2003