ORLA VT
(Oral Reading for Language in Aphasia with Virtual Therapist)

Project Team:
Leora R Cherney 1,2; Edie M Babbitt, 1; Ronald Cole, 3; Sarel Van Vuuren, 3; Rosalind Hurwitz, 1; Jaime Lee, 1, Nattawut Ngampatipatpong, 3

1 Center for Aphasia Research, Rehabilitation Institute of Chicago;
2 Northwestern University, Feinberg School of Medicine;
3 University of Colorado, Boulder.

Project Overview:

ORLA, developed by Dr. Leora Cherney at the Center for Aphasia Research, Rehabilitation Institute of Chicago is one of three projects underway at CSLR that aim to develop computer-based virtual therapist programs that individuals with aphasia can use independently to recover or relearn speech and language.

Aphasia is an acquired multi-modality disturbance of language, resulting from focal damage to portions of the brain, typically within the left cerebral hemisphere, that are responsible for language. The disorder impairs, in varying degrees, the understanding and expression of oral language, as well as reading and writing. People with aphasia experience communication problems that have significant impact on their daily lives; those affected by aphasia report social isolation, loneliness, loss of autonomy, restricted activities, role changes, and stigmatization.

In the United States, the acute stage of aphasia is the locus of most intervention, although research has shown that individuals who receive frequent treatment beyond the period of spontaneous recovery also benefit from treatment. Unfortunately, the shrinking health care dollar continues to limit available services during the acute stage of aphasia, and treatment delivered to patients with chronic aphasia (beyond six months after onset) is seldom reimbursable. There is thus a clear need for innovative and effective ways to deliver much needed treatment to individuals with aphasia who are beyond the acute stage of this disability. Computer treatment may be a cost-effective way of providing such help, thereby meeting the needs of the growing numbers of individuals with chronic aphasia.

Oral Reading for Language in Aphasia (ORLA)

ORLA involves repeated practice reading aloud sentences with a clinician (Cherney et al., 1986; Cherney, 1995). ORLA was developed to improve reading comprehension in individuals with aphasia by providing practice in the phonological and semantic reading routes. Interestingly, the earliest studies of ORLA indicated that individuals improved not only in reading comprehension, but also in other modalities, including oral expression, auditory comprehension, and written expression (Cherney et al., 1986, 1995). Several explanations for the cross-modal generalization have been suggested, and these may be related to other features of the ORLA technique (Cherney et al, 2004). For example, ORLA focuses on connected discourse rather than single words, permitting the modeling of more natural rhythm and intonations. ORLA is also consistent with principles of learning theory, such as active participation by the learner, repetitive practice in the overlearning of skills, and use of meaningful materials that are graded in difficulty.

ORLA has four levels of treatment based on length and reading level:

Level 1. Simple 3-5 word sentences at a first grade reading level;

Level 2. 8-12 words that may be single sentences or two short sentences, at a third grade reading level;

Level 3. 15-30 words, divided into 2-3 sentences, at a sixth grade reading level;

Level 4. 50-100 words comprising a 4-6 sentence simple paragraph, also at a sixth grade reading level.

The graded nature of ORLA makes it appropriate for individuals who present with a broad range of aphasia severities.

Because of its repetitive nature, ORLA may be undertaken relatively independently. Furthermore, since the steps of ORLA are carefully delineated and the tasks are incremental and graded, ORLA had the potential for being administered successfully via computer.

An Initial ORLA Computer Treatment

As part of a NIDRR-funded grant (H133G010098 - PI, Cherney) to investigate the cost-effectiveness of a computerized version of the ORLA, 25 individuals with chronic nonfluent aphasia each received 24 one-hour sessions of ORLA treatment, typically twice a week. These 25 subjects were right handed, with at least a 12th grade education. Age at time of stroke onset ranged from 25.2 years to 80.36 years, and age at time of initial testing ranged from 35.18 years to 81.65 years. A delayed treatment design was utilized; therefore all subjects received treatment following a period of no-treatment. ORLA treatment results were promising, despite the low intensity of the scheduled treatment in this study. On our primary outcome measure, the Western Aphasia Battery Aphasia Quotient (AQ), subjects achieved an average increase of 3.4 (SD= 4.5) points following the low-intensity ORLA treatment as compared to a mean difference of -0.36 (SD= 3.1) AQ points during a delayed treatment control period (Cherney et al., 2005).This study also confirmed the cross-modality improvements but showed that severity of aphasia influenced which modalities improved most.

In this same study, a computer version of ORLA was compared to ORLA treatment delivered by a speech-language pathologist. Although improvements were made on the computer version, these improvements were smaller than those achieved with the speech-language pathologist. A possible reason for this difference was that the patient's receiving the computer ORLA could not see the clinician's face, and thus could not make use of the visual motor information from the lips, tongue and lower face that improve speech production.

Prior treatment studies support this hypothesis, indicating that aphasic individuals, particularly those with nonfluent aphasia and apraxia of speech, benefit from this sensory information. Therefore, in a second NIDRR sponsored study, described below, the ORAL VT program was developed, enabling us to investigate computer-based administration of ORLA using a lifelike computer character that produces accurate visual speech accompanied by natural movements of the head and face.

ORLA Virtual Therapist System (ORLA VT)

With support from another NIDRR grant (H133G040269 - PI, Cherney), we have developed an ORLA Virtual Therapist program to improve the language skills in individuals with nonfluent aphasia. ORLA VT uses a multi-modality stimulation approach that involves several steps including listening to a sentence, tapping along with the rhythm of the sentence, and repeated practice saying the sentence together with the VT and then independently.

Development of the system occurred over a multi-year period in a collaboration between researchers at the Center for Spoken Language Research (CSLR) and Leora Cherney and her colleagues at the Rehabilitation Institute of Chicago (RIC). Development work was done at CSLR starting with an initial design specification that codified the original ORLA treatment, followed by development-and-test cycles in which the rules governing the behavior and interface of ORLA VT system were refined based on feedback from Cherney and colleagues. In addition to developing the treatment paradigm, simple logging and data browsing functionalities were added to facilitate review of patient performance. Finally, the completed prototype system was evaluated at RIC. All functionalities as developed assumed all data to be stored on the client PC only.

Description of system: During the treatment session, the person with aphasia repeatedly reads aloud sentences and paragraphs, first together with the virtual therapist, and then independently. The main features of the therapy are as follows:

1. A stimulus sentence or paragraph appears on the screen. The VT, using a digitized female voice (rather than synthesized speech) with a standard mid-Western American accent, reads aloud to the patient, articulating carefully. The VT is visible to the patient. As each word is read, it is highlighted on the screen.
2. The VT reads aloud to the patient again. Each word is highlighted on the screen and the patient points to each word as it is read.
3. The VT reads the sentence aloud again. This time, the patient points to each word as it is read and highlighted, and attempts to read each word aloud in unison with the VT.
4. Step 3 is repeated. The sound of the VT fades out at the end of the sentence to allow the patient to read the words aloud independently.
5. For each line or sentence, the VT states a word that the patient must then identify. Words may be content words (e.g., nouns, verbs) or function words (e.g., pronouns, prepositions, conjunctions).
6. For each line or sentence, the VT highlights a word for the patient to read aloud. Both content and function words are selected.
7. The patient reads the whole sentence aloud again in unison with the VT.
   

Responses are not forced or corrected; rather, correct responses are modeled while error responses are followed by further stimulation. Stimulation type therapy is the speech and language treatment approach that has been most widely evaluated in treatment efficacy studies, is consistent with principles of learning theory, and is considered representative of the type of treatment that clinicians are actually using in clinical practice.

Learning is graded, with four levels of stimuli based on length and reading level. Determination of the initial level of stimulus is made during the first evaluation session and depends on the Comprehension subscore of the Western Aphasia Battery and the oral reading fluency performance on the GORT-4 (Wiederholt et al., 2001). The first two levels incorporate different stimulus rates (25-35 words per minute and 55-65 words per minute), so that learning within those levels can be gradually increased as progress dictates.

Level 1. Simple 3-5 word sentences at a first grade reading level (e.g., Do you know that man?).

Level 2. 8-12 words that may be single sentences or two short simple sentences, at a third grade reading level (e.g. She bought a new dress on sale for twenty dollars).

Level 3. 15-30 words, divided into 2-3 sentences, at a sixth grade reading level.

Level 4. 50-100 words that comprise 4-6 sentence simple paragraphs; at a sixth grade reading level.

Figure 2 shows screen images from the ORLA VT system. Figure 2a shows the patient receiving treatment following the steps outlined above. Patient data are logged for subsequent browsing by the therapist. Figure 2b shows a display of patient data (which include timing, contextual and audio information) that is easy to navigate. Sentence stimuli were designed following the ORLA protocol outlined above. Figure 2c shows an authoring environment that the therapist used to design patient appropriate stimuli. Finally, figure 2d shows a recording environment for recording these sentences in the voice and at the rate of speaking to be used by the VT.

Figure 2. Selected screen images from the ORLA virtual therapy system
showing various components of the program

Speech data: In preliminary studies, for level 1 sentences, subjects were "talking" about 33% of the time and level 2 sentences, subjects were talking about 50% of the time.

Preliminary Treatment Results: In an ongoing study using ORAL VT, 13 individuals with chronic nonfluent aphasia have used the updated computerized ORLA program for a period of 6 weeks each. All subjects were provided with a laptop computer and practiced at home independently for either 4 hours per week (7 subjects) or 10 hours per week (6 subjects); weekly visits with the speech-language pathologist occurred only to check compliance and to take weekly language probe measures. The mean (SD) change on the Western Aphasia Battery Quotient (AQ) score for the 10 hrs group was 6.4 (5.76) and the 4 hrs group was 5.3 (6.9). These changes compare favorably with the change of only -0.36 (SD= 3.1) AQ points during the delayed no-treatment control period for 25 subjects participating in the previous study. Previous studies have considered a change of 5 points on the WAB Aphasia Quotient (AQ) to be clinically significant (Katz & Wertz, 1997). Data from these 13 subjects provide further support for the efficacy of ORLA.

Further information about ORLA VT can be found in the ORLA VT poster presented at the Poster presented at 2006 ACRM-ASNR Joint Educational Conference, September 27 - October 1, 2006, Boston, Massachusetts

References:

Cherney, L., Merbitz, C. and Grip, J. (1986). Efficacy of oral reading in aphasia treatment outcome. Rehabilitation Literature, 112-119.

Cherney, L. R. (1995). Efficacy of oral reading in the treatment of two patients with chronic Broca's aphasia. Topics in Stroke Rehabilitation, 2(1), 57-67.

Cherney, L. R., Babbitt, E. M., & Oldani, J. (2004). Cross-Modal Improvements During Choral Reading: Case Studies. Presented at the Clinical Aphasiology Conference, Park City, Utah, May, 2004.

Cherney, L.R., Babbitt, E., Oldani, J., Semik, P. (2005). Efficacy of Repeated Choral Reading for Individuals with Chronic Nonfluent Aphasia. Paper presented at the Clinical Aphasiology Conference, Sanibel, FL. June, 2005.

Cherney, L.R., Babbitt, E.,Cole,, R. A., Van Vuuren, S., Hurwitz, R., & Ngampatipatpong, N. (2006). Computer Treatment for Aphasia: Efficacy and Treatment Intensity. Presented at the Annual Education Conference of the American Congress of Rehabilitation Medicine, Boston MA, Sept. 2006.