Predicting AAC Communication Rate

Barry Romich, P.E.1,2,3 , Katya Hill, Ph.D.1,2 and Kate Graham, BA2
1AAC Institute; 2University of Pittsburgh; 3Prentke Romich Company


Communication rate is the single parameter of communication performance for which individuals who use augmentative and alternative communication (AAC) are forced to operate at a significantly lower level than naturally speaking individuals.  Therefore, a major focus of AAC evidence-based practice is optimizing communication rate.  A method is presented for predicting the communication rate of individuals who rely on AAC.  The method is based on 1) the measurement of selection rate of which the individual is capable, 2) reference to the rate indices of the various language representation methods supported by the AAC application program, and 3) relative use of language representation methods.  Example data are reported, demonstrating the efficacy of the method.  Results include the availability of a new tool to support evidence-based practice to optimize the communication performance of individuals who use AAC.


AAC, communication rate, performance, augmentative


The field of AAC addresses the communication needs of people with significant speech disability.  For most people who use AAC, communication rate is far slower than that of naturally speaking individuals.  Communication rate is an important aspect of communication performance which in turn affects the life experience of this population.

AAC professionals strive to maximize communication rate through evidence-based practice service delivery.  Automated language activity monitoring is commonly used for the collection of AAC language samples which include time and content data.  Language samples are analyzed and quantitative communication performance summary measures are calculated using standardized methods and automated means.  Factors that influence communication rate have been identified (1, 2).  They include selection rate (bits per second) (3) and choice and training in the use of AAC application programs.  A rate index (words per bit) has been defined to represent the latter of these factors (4).


AAC application programs frequently employ multiple language representation methods (LRMs).  Further, both the selection rate and the rate index can be a function of the individual LRMs being used within an application program.  Therefore, systematic and evidence-base decision making in the choice and use of AAC application programs that optimize communication performance is uncommon.  Logging provides an accurate method to measure the parameters of communication rate (1, 5).  AAC practitioners need tools and evidence on which to base decisions on AAC system selection and intervention services.


The formula for calculating communication rate (CR) in words per minute based on selection rate (SR) in bits per second and rate index (RI) in words per bit is the following:

CR = SR x RI x 60

Thus, if selection rate for the individual is measured, and the demonstrated rate index for the application program is known, the communication rate potential for the individual using that system can be calculated.

When an application program uses multiple language representation methods, the formula for calculating communication rate can be expanded to accommodate this situation based on the percentage of use of each method.  In the following formula, LRMx represents the various methods being used and the %LRM factors sum to 100%.

CR = ((SR x RI x%)LRM1 + (SR x RI x %)LRM2 + ...

+ (SR x RI x %)LRMx )) x 60


To test the premise of this paper, a language sample from an individual identified as a competent augmented communicator was analyzed to generate selection rate and rate index evidence for each language representation method used.  The individual, an adult with quadriplegic cerebral palsy and normal cognitive function, was using the Unity 128 AAC application program on a Pathfinder device.  The selection method used was manual keyboard direct selection.  The language sample collection protocol was an interview.  This evidence is presented here.

Table 1: Reference evidence on LRM Selection Rate, Rate Index and % Use

Language Representation Method (Unity 128 on Pathfinder)
Selection Rate (bits per second) Rate Index (words per bit) Relative LRM Use (%)
Single Meaning Pictures
Word Prediction
Semantic Compaction

Using the above evidence, communication rate was predicted for five individuals of similar characterization.  This calculation was based on selection rate measured for those individuals using the spelling LRM.  Actual communication rate was measured.  Average predicted communication rate relative to average actual communication rate was 110%, with a range of 87% to 127%.


Use of this method depends on three primary factors.  First is the measurement or estimation of the selection rate of which the individual is capable.  Selection rate may be different for the different language representation methods used.  Methods of measuring selection rate have been defined, one of which is based on spelling.  The relative relationships to be expected among various LRMs have yet to be established, but may be useful in estimating selection rate for various methods if the selection rate for one method is known.  Additional work is needed in this area.

The second factor is the availability of rate index evidence for the application program(s) being considered.  Similarly, rate index may be different for the different language representation methods used and use of LRMs will be a function of vocabulary used.  A database of demonstrated rate indices, based on a group of competent users, is needed.  Such a resource would need to provide evidence for each LRM used with each AAC application program being considered.  Such a resource can be developed based on demonstrated performance of competent users.

The third factor is the relative use of the various LRMs supported by the AAC application program.  Again, this factor will be somewhat specific to the individual and will be a function of use of core and extended vocabulary.  However, some evidence exists of the consistency in relative LRM usage by competent communicators using specific AAC application programs.  This can be used to estimate LRM use for an individual.  The above referenced database can include percentage use for competent individuals.

Use of methods for predicting communication rate for individuals who use AAC can be a powerful tool for assuring that communication performance is optimized.


  1. Hill, K., Holko, R., & Romich, B. (2001).  AAC performance: The elements of communication rate.  Poster presented at the American speech-language-hearing (ASHA) Annual Convention. New Orleans, Louisiana.  November 15-17.
  2. Smith, L. E., Higginbotham, D. J., Lesher, G. W., Moulton, B., & Mathy, P. (2006).  The development of an automated method for analyzing communication rate in augmentative and alternative communication.  Assistive Technology 18, 1. 107-121.
  3. Hill, K., Ramachandran, P., & Romich, B. (2001). Measuring AAC system selection rate. Poster presented at the Annual American speech-language-hearing (ASHA) Annual Convention. New Orleans, Louisiana.  November 15-17.
  4. Hill, K. and Romich, B. (2002). A Rate Index for Augmentative and Alternative Communication.  International Journal of Speech Technology 5, 57-64, 2002.
  5. Lesher, G. W., Moulton, B. J., Rinkus, G. J., & Higginbotham, D. J. (2000). Logging and analysis of augmentative communication.  Proceedings of the RESNA 2000 Annual Conference (pp. 82-85), Orlando, FL. Arlington, VA: RESNA Press.


Barry Romich, P.E.
Chief Operating Officer
AAC Institute
1000 Killarney Drive
Pittsburgh, PA 15234
330-262-1984  x211