Printer Friendly

Treatment of chronic aphasia with errorless learning procedures: a direct replication.

Single-subject experimental designs are often used to measure the effects of aphasia treatment. However, direct and systematic replications (Barlow & Hersen, 1984; Sidman, 1960) of original studies (e.g., Bardin Ayers, Potter & McDearmon, 1975) are uncommon or nonexistent in the aphasia literature (Basso & Caporali, 2004; Siguroardottir & Sighvatsson, 2003, 2005, and 2006). Perhaps one of the reasons for this is that language treatments have not been well-defined (e.g., Byng & Black, 1995; Robey, 1998). Most individual treatment studies have included poorly defined independent variables, thereby obscuring internal validity and rendering replications impossible (Greener, Enderby and Whurr, 2000; Jordan & Hillis, 2006). Definitions and operationalization of independent (treatment) variables and specification of treatment procedures is a necessary prerequisite for experimental replication. Replication of therapeutic effects is a prerequisite for advancement in any therapeutic field. Only with direct and systematic replications of single-subject experimental studies is it possible to ultimately determine which type of therapy is effective for what type of patient with what type of etiology and symptoms and in which particular situation (Barlow & Hersen, 1984; Hayes, Barlow, & Nelson-Gray, 1999). This is especially true when considering a small number of clients with the same etiological classification but with heterogeneous profiles of specific symptoms. This is the case in the field of aphasia because a clear relation between specific symptoms and underlying deficits has not yet been established despite attempts to do so with the newest technology (Hillis, 1989; Thompson, 1996).

Why specific interventions affect the performances of persons with aphasia and how different types of interventions affect those performances remain unknown. Treatment variables that can be linked directly to potential causes of specific symptoms remain undeveloped (Greener, 2000), and the existing research (mostly group studies) does not permit a clear analysis of intervention effects. Group studies are limited because they only provide pooled group averages and no specific information at the individual level. For these reasons, single subject designs are a preferred strategy for studying treatment effects for individuals with aphasia. These designs offer good control over participant variables such as age, gender, type of disability, and others. They also allow the researcher to observe carefully any changes that may occur in a participant's performance during the treatment phase. For example, if indicated by the data, the treatment variables can be adapted to enhance the participant's performance (Siguroardottir & Sighvatsson, 2006). By using well defined treatment variables, this approach also allows for systematic replications (Barlow & Hersen, 1984; Sidman. 1960).

When developing treatments to support patients with Broca's aphasia, it is important to consider the different features of the condition and to focus on how a treatment may affect a particular feature. For example, one patient may have special difficulties naming while another may have problems only when reading. In fact the symptom profiles of patients are often quite unique and they change before the condition gets to be chronic (Cherney & Robey, 2001). Furthermore, when a treatment is being developed it is important that all treatment variables be clearly defined so that the same treatment can be assessed for its effectiveness across different patients with different symptoms.

Single subject research design has been widely used in the field of applied behavior analysis, particularly for evaluating the effects of treatments based on operant learning principles. Treatment based on these principles have been used with efficiency, effectiveness, and efficacy for teaching skills to people with learning disabilities, brain damage, psychiatric disorders and other impairments (for extensive review see Austin and Carr, 2000). Within this framework, symptoms of impairment are viewed as behavioral deficits that can be ameliorated by teaching behaviors that a person is lacking. Treatments that are based on operant procedures and stimulus control are usually well defined, which makes this approach an attractive model for therapists seeking effective, efficient, and efficacious treatments for aphasia. In fact, Austin and Carr (2000) suggest that this framework offers a parsimonious approach for aphasia treatment.

Unfortunately, very few researchers in the field of aphasia have studied the effects of treatment based on operant conditioning or stimulus control procedures (Bardin Ayers, 1975; Doyle, Goldstein, Bourgeois & Nakles, 1989). Presently, very few studies of the effects of operant conditioning on symptoms of aphasia are found in the literature. Bardin Ayers (1975) monitored the error rate of four persons with aphasia whose onset of symptoms ranged from 1 month to 10 years. The participants were asked to work on different tasks depending on their main deficit as determined by test results (the Minnesota Test for Differential Diagnosis of Aphasia (Schuell & Sefer, 1973). Their performances were monitored and a check was made for each correct response. This type of simple performance feedback resulted in a considerable decrease of the error rate. When participants took the test again after 17 sessions over a period of 11 weeks their scores had improved significantly. The results were homogeneous for all four participants, suggesting that variations in the length of time since the onset of symptoms did not seem to affect the results.

Doyle et al. (1989) worked with four participants with aphasia whose onset of symptoms ranged from 29 to 195 months. The dependent variable was requesting information within 20 seconds of being presented with a topic. If no request occurred within 20 seconds the trainers provided a prompt. If the prompt did not evoke a request for information within 20 seconds, then a request was modeled for the participant to imitate. Adequate requests were praised and consequated by the provision of the requested information. The number of requests and their content were monitored. All participants' performances improved considerably. Generalization of requesting was measured by exposing participants to strangers (unfamiliar conversational partners) of whom they could ask questions. Results showed that the participants' requests directed to trainers were better than those directed to strangers. However, the questions directed to strangers were similar to those of a comparison group involving individuals without disabilities. Training effects had not deteriorated after 12 weeks. Unfortunately, there was no evidence of generalization to themes not used during training.

More recently Siguroardottir and her students conducted a series of studies in which individuals with chronic aphasia were to perform a number of verbal tasks successfully, such as naming pictured items, producing sentences, and producing sequences (1999, 2000, 2003, 2005, 2006, 2007). These authors based their treatment on operant conditioning and stimulus control procedures (also known as "errorless training procedures"), using prompts and positive reinforcement in the training phase. The results have been promising, with the performance of the participants improving significantly on all tasks and all of the participants reaching 100% independent correct performance on at least one task.

Sveinsdottir and Siguroardottir (1999) worked with an elderly man with Broca's aphasia, three years post-onset. A multiple baseline across behaviors was used to evaluate the effects of treatment on four conversational behaviors, including a) the rate of correct responding to simple yes/no questions, b) clarity of speech, c) rate of eye contact, and d) initiation of conversation. The intervention was conducted during only 20 sessions over four weeks. It consisted of four elements: a) instructions for each task; b) praise and other social reinforcements for correct performance on 100% of all training trials; c) verbal performance feedback at the end of each session; and d) visual performance feedback in the form of weekly graphs. Clinically significant changes were observed on all dependent variables. Especially striking was the change in performance in the initiation of conversation task, which progressed from almost nothing to 35 times during a 20- minute session as a result of training. At 16 weeks follow-up, measures showed that training effects maintained and the participant's performances had improved, perhaps because family members, for whom the intervention was explained and modeled, continued to use the main treatment variables after formal training had ceased.

Porsteinsson and Siguroardottir (2007) used backward chaining to teach a 55-year old woman with chronic aphasia to read compound Icelandic words. Each compound word was divisible into three component words. Modeling and differential reinforcement were used during each trial of the training phase. The participant made considerable progress. In addition, training effects generalized to two contexts: a) labeling pictured objects whose names corresponded to the trained words, and b) labeling pictured objects whose names were not targeted during the intervention phase.

Siguroardottir and Sighvatsson (2003, 2005, and 2006) worked with four people with Broca's aphasia whose onset of symptoms ranged from 1? to 4 ? years. Multiple-baseline across behaviors and AB designs were used to evaluate the effects of treatment on a number of conversational behaviors, including a) rate of correct responding to questions, b) naming people and objects, and c) making sentences. Clinically significant changes were observed on a number of dependent variables with all four participants. However, the systematic search for equally important treatment variables (e.g., the generality and limitations of treatment effects) remained an open question. Additionally direct and systematic replications of single-subject experiments are needed more generally (Barlow, Nock & Hersen, 2009; Sidman, 1960). The study reported here is one attempt to initiate this process. It is a direct replication of the Siguroardottir and Sighvatsson (2003, 2006) study for the purpose of measuring the generality of effects of a treatment.



Three men, ages 51 to 63, participated in this study. All had suffered a single cerebrovascular accident, and all had been originally diagnosed as having severe Broca's aphasia by the head speech-language pathologists (SLPs) at Iceland's University Hospital. Twelve to 62 months had passed since the incident that led to the diagnosis; thus, all three presented with chronic aphasia. The status of each participant at the beginning of the study is summarized in Table 1.
Table 1. Participants age, gender, and time since onset of

Participant Age Gender Months Type of
 since Aphasia

1 63 Male 36 Severe

2 54 Male 12 Severe

3 51 Male 62 Severe


Baseline, training, and generalization sessions were conducted in each participant's home.


Baseline and training phases. The visual stimuli consisted of color photographs that were glued on to the center of 21 x 15 cm. index cards and then laminated. All 130 pictures were used with all participants.

Generalization phase. Fifteen color photographs (photo cue cards) (Kerr, 1979; 1985) were used. Each card was 15 x 10 cm. in size. These cards were similar to those used in training. All cards were used with all participants.


Dependent variables (target skills). The dependent variables consisted of target skills that were selected for each participant based on consultation with their SLP and spouses about performances that proved difficult to them. Table 2, below, summarizes the specific targets selected for each participant.
Table 2. Dependent variables selected for each participant

Participant Dependent variables (Target Skills)

1 a. Verbalizing sentences when shown a
 picture and asked "What is happening in
 the picture?" or "What can you see in
 this picture?"
 b. Naming a sequence when given a request
 like "Name the days of the week"

2 a. Answering "Yes" or "No" to questions
 like "is it Monday today?"
 b. Naming digits when shown a digit on a

3 a. Naming familiar people (e.g., family
 members, celebrities or politicians)
 b. Naming objects commonly found in daily
 living (e.g., household items, a car
 c. Naming letters from the alphabet.

Scoring. Participants' responses were scored as correct, incorrect, or no answer. Each trial consisted of up to two response opportunities. During each response opportunity, the participants were given at least 10 seconds to respond. If a participant made a correct or incorrect response during the first opportunity, the response was scored as "correct" or "incorrect," respectively. If a participant made no response during the first opportunity, the question or request was repeated and a second opportunity was presented. If the participant did not respond during the second opportunity, the experimenter scored the response as "no answer" and initiated the next trial. If the participant indicated a refusal to respond or indicated that he had no answer, the response was scored as "no answer" and the next trial was initiated.

Dependent measures. The main dependent measure in all tasks was the number of correct responses out of trials given.


The intervention consisted of seven parts: a) giving standardized instructions for each task, b) setting the occasion for the target behavior, c) providing encouragement after each response, d) correcting errors after scoring by demonstrating the correct response and requesting imitation, e) rehearsal (at least 3 opportunities), f) praising appropriate responses, and g) weekly feedback in the form of graphs illustrating the rate of correct performance during the previous week.


A multiple-baseline across tasks design was used to assess the effects of the treatment on the the dependent variables (target skills) selected for each participant. However, if the treatment did not lead to improvement for a particular skill over 3 consecutive sessions, the skill was broken down into smaller steps, taught through backward chaining, and assessed within an AB design that was not part of the multiple baseline. For example, if the task was to name digits but measures showed that the participant could only name the digit when it was modeled by the experimenter, then digit naming was broken down into smaller steps and taught through backward chaining. Those data became part of an AB design and not part of the multiple baseline.

Baseline. Baseline measures of all dependent variables selected for each participant commenced simultaneously. Measures were taken daily on five consecutive weekdays. Each session lasted from 25 to 45 minutes. At least five baseline measures were taken before the intervention began on the first dependent variable.

Intervention. The intervention lasted for 6 months. Sessions were held daily during weekdays and lasted from 30 to 60 minutes, depending on the type of tasks selected with each participant and on their performances in those tasks.

Variations in the intervention principally involved the amount of help (prompt level) provided to a participant. Three prompt levels were used, each providing successively less help. Level 3 provided the most help, and it was used in the beginning stages for all participants in all tasks. At this level, the complete target response was modeled. For example, if the task was to verbalize sentences, the experimenter modeled a sentence and asked the participant to imitate it. When a participant performed with 100% accuracy (prompted or unprompted) over four consecutive sessions, the amount of help was reduced to Level 2. This level initiated prompt fading. Specifically, after presenting the participant with a task, the experimenter waited for 10 seconds. If the participant did not verbalize the target sentence, the experimenter provided a partial model (e.g., said the first two words of the target sentence) and waited for the participant to finish the sentence. When a participant performed at 100% accuracy on three consecutive sessions at Level 2, the amount of prompting was decreased further to Level 1. At this level, participants received an indirect prompt. For example, when asked to verbalize a sentence, the experimenter pointed to an item (e.g., a house) on the stimulus picture and asked, "What can you tell me about this?

Once a participant's performance was stable at Level 1 prompting, the rate of prompting was thinned and randomized systematically. Initially, it was lowered from 100% (i.e., prompting on 10 of 10 trials) to 67% probability (7 of 10 trials). If a participant's performance did not deteriorate, or deteriorated on only one response at this prompt rate, then the rate was lowered further to 33% (3 of 10 trials). When the prompt rate was lower than 100%, the specific trials to be prompted were selected randomly and marked on the score sheet prior to instruction. These intermediate prompting steps were used to fade prompting while maintaining correct task performance in order to prevent the loss of stimulus control.

Recycling. If lowering the prompt rate led to a deterioration in performance two sessions in a row, the prompt rate was returned to the previous effective level. For example, if a participant performed well at a 67% probability of prompts per trial but his performance deteriorated at 33%, then the 67% prompt rate was restored.

Generalization. Generalization was assessed during the last four weeks of the study, three times per week. When generalization was assessed, training was shortened to a maximum of 30 minutes or omitted altogether.


Reliability was scored both with regard to the dependent variables (response accuracy) and independent variables (i.e., prompt levels). With one exception, reliability was assessed for 30% of all baseline measures. In the case of one participant, no reliability measures were taken during baseline measures during one task due to experimenter error. However, reliability was assessed for up to 50% of all other baseline measures.

During intervention, reliability was measured in 18.5% to 25% of sessions across participants and tasks. During generalization, reliability was measured in 25 to 30% of sessions across participants and tasks. Table 3 shows mean reliability for each participant in each task during baseline, training and generalization. Table 4 shows mean reliability for measures of correct prompting with each participant in each task in both the direct replication study and the extended study.
Table 3. Mean reliability measures for main dependent variable
during baseline, training, and generalization in each task with
each participant.

 Tasks 1 2 3

 Sequencing N/A - 93%

 Sentences 90% - -

 Numbers - 94% -

Baseline Yes/No - 98% -

 Naming - 97.50% -

 Naming - - 95%

 Naming - - 94%

 Letters - - 98%

 Sequencing 100% - 95.84%

 Sentences 91.10% - -

 Numbers - 95.29% -

Training Yes/No - 98.13% -

 Naming - 98.63% -

 Naming - - 96%

 Naming - - 96.25%

 Letters - - 99.33%

 Sequencing 100% - N/A

 Sentences 100% - -

Generalization Numbers - 86.40% -

 Yes/No - 97.40% -

 Naming - 94.80% -

 Naming - - N/A

 Naming - - N/A

 Letters - - N/A

N/A = not applicable, measures not taken. It was not possible to
take generalization measures because the participant never reached
that point

Table 4. Mean reliability measures for use of correct prompt.


Correct Prompt Use 1 2 3

Sequencing 91.75% - -

Sentences 100% - -

Numbers - 100% -

Yes/No Answers - 100% -

Naming objects (simple stimuli) - 100% -

Naming objects - - 99.50%

Naming people - - 100%

Letters - - 97.11%


The results of the current study replicated the results of Siguroardottir & Sighvatsson (2006) partially. The treatment had clinically significant effects on the performances of participants 1 and 2. However, the performances of participant 3 could not be brought under the targeted stimulus control. An extended intervention was developed to address his needs, as reported in a second experiment, described below. The remaining portions of this results section pertain only to the performances of participants 1 and 2.


Performances of participants during baseline and training can be seen in Figures 1 to 5. In general, the intervention had clinically significant effects on the performances of two participants across tasks. Participant 1 performed well on both the sequencing task and in verbalizing sentences. His performance was better on sequencing, where he reached the point of perfect performance without any help from the experimenter. His performance in verbalizing sentences showed good progress, but it leveled off and remained under the control of the experimenter's help. Thus, his performance was very comparable to the performance of participant 1 in Siguroardottir & Sighvatsson's original study, but it was not as impressive.






The performance of participant 2 on naming numbers was unstable but he did make some progress as can be seen on Figure 3. His performance on answering correctly "yes" or "no" to questions was good. This participant had regularly answered "yes" and "no" to the same question in the year previous to participating in this study. That is, he answered either in that order or in the opposite order (i.e., said "no" first, then said "yes"), so his conversational partners were never sure of which answer he intended to give or whether his responses even related to the questions he had been asked. He had been trained to use manual signing before participating in this experiment. However, as in the case of his verbal responses, he sometimes used both signs ("yes" and "no") in sequence, leaving his communication partners confused. For these reasons, it was important to teach Participant 2 to use only one verbal answer per question. In this study, he was trained to use oral "yes" or "no" responses only and he was discouraged from using the manual signs during training. As can be seen in figure 4, the first intervention was not successful. Therefore, the intervention was modified, but it failed again to affect his performance. Hence, a second intervention modification was developed, which was successful. This intervention included prompt Level 3 with three components: a) instructions about tongue position for "yes" or "no", b) modeling the correct response without sound, and c) saying the correct answer slowly, loudly, and with exaggerated mouth movements. When his performance was correct three times in a row in 100% of 10 opportunities, he was offered 30 opportunities to respond instead of 10. At the same time, the prompt level was decreased to level 2, such that the instructions about the position of the tongue were skipped, the correct answer was modeled and the mouth movements were still exaggerated. However, his performance became unstable at Level 2, so prompt Level 3 was resumed. The experimenter then followed the recycling-criteria used throughout the experiment. Using this strategy, the performance of Participant 2 did improve. His manual signing was monitored and figure 5 shows his performance in manual signing "yes" and "no" as his oral "yes" and "no" answering was trained. It is clear that his performance in manual signing declined at the same time as his oral performance improved. This is interesting. There was no other intervention attempted to reduce his use of manual signing other than instructions not to sign at the beginning of the training.

During measures of generalization, participant 1 showed stable generalization on both tasks (i.e., sequencing and verbalizing sentences). Participant 2 on the other hand showed little or no generalization across situations or settings. Generalization was most clear when the person he knew best, of those available to test generalization across trainers, was the one who worked with him through the tasks that measured generalization. This person had often taken reliability measures, thus, Participant 2 had often been exposed to him while working through the tasks.


This study was a direct replication of Siguroardottir and Sighvatsson (2003, 2006). The results were replicated partially. The effects of a behavioral training method for symptoms of chronic aphasia were demonstrated in two single-subject experiments. One experiment used a multiple-baseline design across behaviors with participant 1. The other experiment used an A-B design with two behaviors with participant 2. The treatment-variables in this study were the same as in 2006 study, and they seem to have had the same effect on the performances of the participants in this study. Participant 1 seems to have been disabled to a similar degree as participant 1 in the Siguroardottir and Sighvatsson (2006) study. However, he appeared less motivated. For example, sometimes he did not answer the door when the experimenter arrived, and sometimes he seemed more interested in the experimenter's company than in working on the task. Also, he was more socially isolated in that he had few friends and family visited seldom. Participant 2, however, was much more disabled than any of the participants in the Siguroardottir and Sighvatsson original study. Nonetheless, the intervention affected his performances in a similar way, although not in the same magnitude. An additional experiment was done to address the problems in other performances of participant 2. An intervention was developed for those more severely disabled performances, and backward chaining was used as the main intervention.

As in the Siguroardottir and Sighvatsson (2006) study, it is not clear from the present study which variable or variables are the most important, sufficient, or necessary to achieve the treatment effects observed. However, this study is an important contribution to the direct replication process needed to establish the effectiveness of interventions through single subject design. The treatment did have clinically significant effects on most of the behaviors selected for intervention for two of three participants. However, the results are not as impressive as those in the 2006 study as seen, for example by the level of generalization attained. The reason for this is not clear. Treatment-variables were the same. Intervention lasted almost as long as in the original study, and the experimenters were the same. The only one variable that was known to be different was the level of disability of the participants. Two participants in this study had more severe types of aphasia, their neurological injuries were much more severe. For example, in the course of the experiment it became clear that not all participants could imitate oral behaviors modelled by the experimenter. Perhaps inability to imitate verbal responses will limit the effects of the treatment described here and in the original study. However, it is still too early for this conclusion to be reached, further experimental analysis of treatment variables is necessary.

Since the results of Siguroardottir and Sighvatsson (2003, 2006) were only partly replicated in the study reported above, an additional experiment was undertaken with the objective of bringing the behaviors of two participants under stimulus control. Backward chaining was used as the main intervention in the next experiment reported here. Both participants presented with severe Broca's aphasia including markedly deficient vocal imitation repertoires as described below.


The method used in the extended study involved slight variations from the method used in the study described above. The following information summarizes these variations.


Participants 2 and 3 from the study reported in the beginning of this paper served as participants in this extended study. Please refer to their profiles in Table 1. Neither participant imitated sounds at the beginning of the study. Each could imitate movements of the speech mechanism, bust, only Participant 2 reliably followed instructions to do so.


The following tasks were selected for both participants: a) Naming objects; b) Naming people; c) Naming a routine sequence (e.g., days of the week; months of the year); d) Naming letters. As in the study reported above, each of these targets was selected to lessen the major symptoms of Broca's aphasia.


A multiple-baseline across tasks was used to assess the effects of the treatment on the skills targeted for Participant 3. An A-B design was used to assess the performance of skills targeted for Participant 2 and for one behavior targeted for Participant 3. Assessment began in all assessed variables simultaneously.

Baseline. Measures were taken daily. Sessions lasted for 15 to 45 minutes.

Intervention. Backward chaining was used as the main intervention in the present study instead of the errorless procedures described in the initial study. Three levels of promopting were manipulated. At prompt Level 3 for all tasks, both participants were asked to say the last sound in a target (i.e., the last sound in the word for the name of a person, an object, an item of a sequence, a letter name). Specifically, the experimenter modeled all parts of the target minus the last sound and instructed the participant to produce the last sound. When a participant performed with 100% accuracy over four consecutive sessions, the amount of help was reduced to prompt Level 2. At this level, the participant was asked to say the last two sounds in the target word. When a participant performed with 100% accuracy on four consecutive sessions, the amount of help was again decreased to prompt Level 1. At this level, a participant was asked to say the last three sounds in the target word. If a participant failed to say the missing sounds, he was offered the opportunity to rehearse. After scoring the response as incorrect or as "did not respond", three additional steps were followed: The whole word was modeled by the experimenter. The word-final sound (which the participant had failed to say) was exaggerated, and the participant was asked to imitate the word.

Recycling. If a participant's performance deteriorated two or three times in a row across sessions, or if the performance became unstable over several sessions, the level of help was increased in the following session and the criterion for lowering the level of help was doubled.



Performances of participants during baseline and training in the tasks can be seen in Figures 6, 7 and 8. Figures 6 and 7 show performances of participant 3. His performance was unstable, especially in the Naming People and Sequencing tasks. However, in the Naming Objects and Naming Letters tasks, he performed better. In the Naming People task, his performance was variable after intervention started. He produced between 6 and 10 correct answers out of 10 response opportunities. In the Sequencing task, he produced between 1 and 3 correct answers out of 3 response opportunities. That performance was less variable than in the Naming People task. His performance in the Naming Objects task was good, ranging between 10 and 15 correct responses to 15 response opportunities. His poorest performance occurred while sick or injured following a car accident.



Although the difference between Participant 3's performance in baseline and intervention seems clinically significant, it must be noted that his performance never progressed beyond Level 2 prompting in backward chaining on any tasks. When the treatment was terminated, the level of prompting on all tasks was Level 3 backward chaining, as can be seen in figure 6. Many attempts were made to withdraw the level of help with backward chaining in the Naming People, Naming Objects, and Sequencing tasks, but this only resulted in loss of stimulus control. Participant 3 did not imitate the names of objects, persons, or any other word independently. It became apparent, throughout the training that he was also unable to imitate single sounds.

It was noted that if Participant 3 saw the first letter of the word in a correct response he sometimes, though inconsistently, said the name of the target person or object. This observation led to an assessment of the participant's letter naming performance stability. Figure 7 shows the participant's best performance on the Naming Letters task used late in the training phase. It was hypothesized that teaching him to name alphabet letters reliably would help to improve his performance on the other tasks. Unfortunately this hypothesis was not confirmed by the results. Therefore, it was concluded that this intervention was unsuccessful for most tasks for Participant 3. Perhaps his inability to imitate sounds was an important prerequisite for the training to be effective.

Figure 8 shows the performance of participant 2 on the Naming objects task. Specially selected simple pictures of single stimuli were chosen. As can be seen in Figure 8, his performance was stable in the first part of the intervention. However, as the study advanced, his performance deteriorated. The most reasonable explanation for this deterioration is that backward chaining was terminated and prompt Level 3 (Siguroardottir & Sighvatsson, 2006) was provided instead (i.e., hearing the first 3 to 4 sounds in a word to prompt word completion). When he was helped with backward chaining, and thus had only to provide the last sound of the name of the target object, his performance was under good stimulus control, but not when the prompt level was decreased. Participant 2 presents with severe Broca's aphasia and is mostly unable to imitate modeled words (imitated in 1 out of 10 trials during baseline).


In this extended study, a backward chaining procedure was used to treat two participants who presented with severe Broca's aphasia. These individuals were much more disabled than any of the participants in Siguroardottir and Sighvatsson's (2006) study and one of the participants in the replication study reported earlier in this paper (participants 1, see Table1). One of the participants in the extension study was barely able to imitate and the other one was unable to imitate at all. This extension study was conducted because a direct replication study of Siguroardottir and Sighvatsson (2006) was only partially successful. Therefore, backward chaining procedures were used to target stimulus control of the participants' performances when the errorless learning procedures used in the original replication study failed. The performances of both participants improved very slowly. However, both remained prompt dependent, although one improved faster than the other.

Treatment variables were the same as in Siguroardottir and Sighvatsson's (2006) study. However, as the intervention results were unsuccessful, new intervention variables and new target behaviors were selected that were seen as prerequisites to the target behavior. Participant 3's performance on three of the four tasks was unstable and under little or no stimulus control from the intervention. To address his unstable performance late in the treatment, a new task (Naming Letters) was introduced under the hypothesis that reliable letter naming would improve this participant's performance on the other tasks. Unfortunately, that did not happen. Participant 3 and his family were very interested in participating in the study and in the treatment. Unfortunately, this interest and involvement had no effect on the participant's performances.

Participant 2 and his family were also very motivated by the treatment and the study. Due to his severe condition, his ability to imitate and to be understood by others was minimal. That is why backward chaining procedures were used to target Naming objects. His behavior was quite stable when backward chaining procedures were used but soon after prompt Level 3 (described in the first experiment) was initiated his performance no longer remained under the stimulus control. As can be seen from Figure 8 (and from results in the previous study) participant 2 needed intensive prompting to maintain stable performance. Perhaps, after much longer-term training, prompts could have been faded more successfully, but the amount of training provided in the present study did not lead to that achievement.

The results of this and the previous study indicate that a participant's degree of disability may be a factor in the success of errorless procedures on conversational behaviors of people with severe aphasia, at least when training continues for only 5 to 6 months. Of particular concern is the ability to imitate. The studies reported here used a linguistic approach when target behaviors were selected for treatment. Assessments had been conducted with traditional tools used by speech pathologists, those tools do not emphasize a functional analysis of symptoms. A functional analytic approach of speech-language skills may be more fruitful (Esch, LaLonde, and Esch (2010). The effects of a functional approach in the assessment and treatment of aphasia needs to be studied in depth. It may be much more likely to provide answers to questions about appropriate treatment of individual aphasia symptoms.

Further examination of the variables necessary to bring the conversational behaviors of persons with severe chronic aphasia under stimulus control by an intervention based on errorless learning and operant conditioning is already underway.


Austin, J. & Carr, J. C., Eds. (2000). Handbook of Applied Behavior Analysis. Reno, Nevada: Context Press.

Bardin Ayers, S. K., Potter, R. E. & McDermon, J. R. (1975). Using reinforcement therapy and precision teaching techniques with adult aphasics. Journal of Behavior Therapy and Experimental Psychiatry, 6, 301-305.

Barlow, D. H. & Hersen, M. (1984). Single case experimental designs. Strategies for studying behavior change. New York: Pergamon Press.

Basso, A., Caporali, A. (2004). Targeted intervention for multiple language disorders: a case study. Journal of Neurolinguistics, 17(5), 383-401

Byng, S. & Black, M. (1995). What makes a therapy? Some parameters of therapeutic intervention in aphasia. European Journal of Disorders of Communication, 30, 303-316.

Cherney, L.R. & Robey, R.R. (2001). Aphasia Treatment: Recovery, Prognosis, and Clinical Effectiveness. In R. Chapey (ed.), Language Intervention Strategies in Aphasia and Related Neurogenic Communication Disorders (pp. 148-172). Baltimore: Lippincott, Williams & Wilkins.

Doyle, P. J., Goldstein, H., Bourgeois, M. S., & Nakles, K. O. (1989). Facilitating generalized requesting behavior in Broca's aphasia: An experimental analysis of a generalization training procedure. Journal of Applied Behavior Analysis, 22 (2), 157-170.

Esch, B. E., LaLonde, K. B., & Esch, J. W. (2010). Speech and language assessment: A verbal behavior analysis. Journal of Speech-Language Pathology and Applied Behavior Analysis, 5, 166-191.

Greener, J., Enderby, P., Whurr, R. (2000). Speech and language therapy for aphasia following stroke. Cochrane Database of Systematic Reviews. (2):CD000425.

Hayes, S. C., Barlow, D. H. & Nelson-Gray, R. O. (1999). The Scientist Practitioner. New York: Pergamon Press.

Hillis, A. E. (1989). Efficacy and generalization of treatment for aphasic naming errors. Archives of Physical Rehabilitation, 70, 632-636.

Jordan, L.C., Hillis, A.E. (2006) Disorders of speech and language: aphasia, apraxia and dysarthria. Current opinion in neurology, 19(6), 580-585

Kerr, J. Y. K. (1979 & 1985). Photo Cue Cards & More Photo Cue Cards. Texas: ProEd.

Robey, R. R. (1998). A meta-analysis of clinical outcomes in the treatment to aphasia. Journal of Speech, Language, and Hearing Research, 41, 172-187.

Schuell, H. & Sefer, J. W. (1973). Minnesota Test of Differential Diagnosis of Aphasia (MTDDA). Minneapolis, MN: University of Minnesota Press.

Sidman, M. (1960). Minnesota Test of Differential Diagnosis of Aphasia (MTDDA). Minneapolis, MN: University of Minnesota Press.

Siguroardottir, Z. G. & Sighvatsson, M. B. (2003). Treatment of chronic aphasia with errorless learning procedures and operant conditioning. In F.H. Jonsson (Ed.), Rannsoknir i Felagsvisindum s IV (pp. 505-518). Reykjavik: Haskoli Islands. (Conference proceedings).

Siguroardottir, Z.G. & Sighvatsson, M.B. (2005). Treatment of chronic aphasia with errorless learning procedures and operant conditioning: A direct replication. In Ulfar Hauksson, ed. Rannsoknir i felagsvisindum VI. Reykjavik: Felagsvisindastofnun. (Conference proceedings).

Siguroardottir, Z. G. & Sighvatsson, M. B. (2006). Operant conditioning and errorless learning procedures in the treatment of chronic aphasia. International Journal of Psychology, 41, 527-540.

Sveinsdottir, I., & Siguroardottir, Z. G. (1999). Modification of conversation behaviors in a patient with communication dysfunction, typical of Broca's aphasia. Poster session presented at the annual conference of Applied Behavior Analysis-International, Chicago Illinois.

Thompson, C. K. (1996). Linguistic-Specific Sentence Production Treatment for agrammatic Aphasia. Topics in Stroke Rehabilitation, 3, 60-85.

Porsteinsson, H. & Siguroardottir, Z. G. (2007). Backward Chaining Used to teach a Woman with Aphasia to Read Compound Words: A Single Case Study. 6/3 $%$ 2, Issue No. 3, 325.

Avar Porolfsson, (2000). Applied behavior analysis: The effects of training conversational skills in a man with chronic aphasia. Unpublished BA thesis. University of Iceland.

We thankfully acknowledge the Icelandic Government Research Fund (RANNIS) and the University of Iceland Research Fund which very kindly supported the research reported here.



Associate Professor Faculty of psychology University of Iceland Oddi

101 Reykjavik, Iceland Phone: (354) 698-7450


Licenced Psychologist Landsspitalinn-University hospital Psychiatry unit 105 Reykjavik, Iceland Phone: 354) 824-6957

Zuilma Gabriela Siguroardottir - University of Iceland

Magnus Blondahl Sighvatsson - Landsspitalinn-University Hospital
COPYRIGHT 2012 Behavior Analyst Online
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2012 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Siguroardottir, Zuilma Gabriela; Sighvatsson, Magnus Blondahl
Publication:The Journal of Speech-Language Pathology and Applied Behavior Analysis
Article Type:Report
Geographic Code:1U4MN
Date:Aug 1, 2012
Previous Article:Modified stimulus presentation to teach simple discrimination within picture exchange communication system training.
Next Article:Caregiver training in an AAC intervention for severe aphasia.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters