Using Speech and Language Technology to Coach 
Reading 
Patti Price 
Formerly BravoBrava! LLC 
Now PPRICE Speech and Language  
 
1-650-503-7053 
pjp@pprice.com 
 Luc Julia 
BravoBrava! LLC 
32980 AlvaradoNiles Road, Suite 856 
Union City, CA 94587  
1-510-477-0493 
Julia@bravobrava.com 
 
 
  
ABSTRACT 
BravoBrava! is expanding the repertoire of commercial user 
interfaces by incorporating multimodal techniques combining 
traditional point and click interfaces with speech recognition, 
speech synthesis, and gesture recognition.  One of these 
applications is software to help children read. While the child 
reads aloud, the computer keeps the child on track and offers 
feedback when the child has difficulty. The feedback can be as 
subtle as changing the text color for a well articulated phrase or as 
friendly as a cartoon character that talks. The computer is 
infinitely patient and can keep detailed records of the child's 
progress.  The reading software is being commercialized by 
BravoBrava!’s spinoff company, SUP Inc. 
Keywords: Reading, pedagogy 
 
 
Figure 1. Here is an example from a sample story.  The system 
tracks the reading and intervenes when help is needed.  This 
screen shows a sample intervention if the child stumbles on the 
word ‘butterflies’. 
 
 
 
 
1. INTRODUCTION 
Our vision is to use technology to provide a high-quality, low cost 
reading coach that delivers voice-activated reading instruction, 
practice, and assessment over electronic media.  Reading is 
fundamental; it can also be fun. However, about 40% of 
mainstream 4th graders cannot read at the basic level.  One of our 
country’s critical needs is to improve reading performance for all 
children since their future, both individually and together, 
depends on literacy.  Reading level predicts economic 
performance for both individuals and societies.   
Beyond the basics, as reviewed in the recent report of the National 
Reading Panel (2000), engaging children in supported oral 
reading is the most valuable means toward building their reading 
proficiency.  At present, however, the only means of giving 
children such practice is by finding a human adult who will sit 
with them and help them.  However, technology can provide an 
automated reading coach to break through this bottleneck, so that 
every child in every school can get the support that she or he 
needs.  This technology will help reduce the digital divide, and 
provide an unprecedented level of tracking data to leverage 
teachers’ instruction and assessment efforts and to build the next 
generation of intervention techniques. 
2. SUP’S APPROACH 
SUP targets the stage at which children have learned the letter to 
sound rules but are still struggling to gain vocabulary and fluency.  
This stage has sometimes been called the transition from ‘learning 
to read’ to ‘reading to learn’ and comes just after the stage 
requiring explicit pronunciation and vocabulary tutoring, 
provided by Mostow’s Project Listen at CMU and other reading 
software.  At this stage in reading development, language skills 
are usually too poor to make traditional dictionaries much of a 
help.  Dictionaries disrupt the child’s focus on the text being read, 
offer too many definitions, and usually provide definitions that are 
harder for the child to read than the original text.  Instead, SUP 
provides a generalized ‘dictionary’, the Reading Resource, to give 
immediate context specific help when needed, much the way a 
human reading coach might intervene.  “SUPplementing” text 
with the Reading Resource, which includes word definitions, 
sample sentences, graphics and multimedia can create an engaging 
environment for learning.  However, the immediacy and context-
specificity of this resource aims to maintain the child’s focus on 
the material being read.  Becoming engaged in the text itself is the 
real goal of reading.  SUPplementation can make texts accessible 
to children that would otherwise be above their reading level, so 
that the content of the text can draw them into the desire to read.  
Vocabulary and grammatical knowledge grow with experience of 
more words in more contexts. 
 
 
 
Figure 2. The screen shot above a sample after the child has 
asked for a display of how well the selection was read. 
 
SUP has developed a modular architecture that allows for rapid 
reconfiguration: 
 
• The current demonstration uses Microsoft speech 
recognition software.  However, the architecture 
supports the use of other recognizers, and we have 
experimented with others for use on other platforms.   
• The audio outputs can be from recorded waveforms or, 
for maximum flexibility, can use a text to speech 
synthesis system. 
• Any text can flow through the system for reading 
practice. 
 
Although explicit measures, such as a quiz to assess 
comprehension or vocabulary, can be included, several automatic 
measures are important by-products of use of the tool: 
 
• Total number of words read by session, and which were 
fluent or not 
• Words per minute as a function of time 
• Level of the material read as a function of time 
• Number of times intervention of the Reading Resource 
was used 
• Number of times the child had the story read to him/her 
• The actual recordings of what was read 
3. FUTURE DIRECTIONS 
Of course, such software will not help unless it gets into the hands 
of the children who need it.  Therefore, an important strategy of 
the company is to support as many different platforms as possible 
to enable this goal.  The modular architecture of the system 
facilitates transition to various platforms.  Similarly, to be of 
maximal interest to the most children, we need as much appealing 
content as possible.  In the area of human language technology, 
we hope such technology will evolve to automatically provide: 
 
• Context-specific definitions  
• Context-specific and child-specific synonyms 
• Rewriting of complex phrases into simpler ones 
 
Solving these problems completely is a major research project.  
However, we believe that with careful matchmaking between the 
technical possibilities and the real needs of beginning readers that 
progress can be made in the near term. 
 
We hope that experience in the usability trials, just beginning, will 
also help us to plan similar products for adult language learners 
and learners of English as a second language.  Reading is still the 
principal way that people learn information.  In this information 
age, reading well is crucial to the individual and to society. 
 
 
 
 
 
 
 
