音声ブラウザご使用の方向け: SKIP NAVI GOTO NAVI

Web Posted on: August 4, 1998

A multimedia system for home-based delivery of language therapy

Alison Crerar

Department of Computing, Napier University
Edinburgh, Scotland
tel: +44 (0)131 455 4623,
fax: +44 (0)131 455 4552,
e-mail: mac@dcs.napier.ac.uk


1. Summary

Information technology promises new possibilities for those with language disorders. So far more effort has been put into the development of augmentative communication devices, or prostheses, to assist those with limited speech output than into exploring the rehabilitation prospects for those with cognitive deficits resulting in impaired comprehension. Our work has focused on stroke victims with sentence comprehension difficulties. We have shown in a previous study using purpose built software known as 'Microworld for Aphasia', that given appropriately targeted treatment, some stroke victims are capable of significant and durable improvements in their language comprehension, even many years post-onset. Unfortunately, the position in the UK is that only a small amount of language therapy is provided by the National Health Service; this usually happens in the early months following a stroke, after which the client is discharged with the prognosis that further improvement is unlikely. We therefore decided to re-design our software with the aim of making it suitable for home use. A small trial has been conducted to test the usability and efficacy of this multimedia version. Preliminary results suggest that we have achieved a therapeutically useful tool, which requires no trained clinical input during use, and which offers the potential for clients and clinicians to exchange exercise files and progress reports electronically.

| Top |

2. Introduction

'Aphasia' is a general term covering a range of language impairments, both expressive and receptive. Adult aphasia, for example following stroke, is a non-acute condition where more or better therapy is still unlikely to restore victims to their pre-morbid state. In the current financial climate, investment must be seen to result in tangible benefits: unfortunately, in this sense, aphasia therapy is not considered cost effective since improved quality of life is an economic 'intangible'. Nevertheless, if we could demonstrate the clinical effectiveness of computers in terms of better treatment outcomes, or less treatment hours necessary, or a more effective utilisation of clinicians' time, the case for investment in technology would be a much stronger one.

Our work has focused on asyntactic comprehension of written English sentences, a distressing condition whereby sufferers cannot understand simple sentences reliably, even when they can be shown to understand most of the constituent words in isolation. So, for example, given a sentence such as 'The cow kicks the horse', the individual might understand that a cow and a horse are involved, but would be unsure about who is doing the kicking and who is being kicked. In English the meaning can only be derived by knowing the grammatical rules of subject-verb-object word ordering for active sentences. The established way of investigating agrammatic comprehension is by means of 'reversible sentences'; these are sentences where the subject and object can be interchanged with equal plausibility. The reason is that non-reversible sentences such as 'The boy rides the bicycle' and implausible sentences such as 'The dog chases the cat' can be interpreted by intelligent 'guesswork', whereas fully reversible sentences like 'The man calls the woman' cannot and are therefore more diagnostically useful.

As with other aspects of psychometrics, there are two dimensions of performance to consider in determining the level of impairment: speed of processing and accuracy of interpretation. Speed of processing is rarely mentioned in the aphasia literature, but it is crucial to everyday functioning. Using a computer to administer assessments, accurate timing data can be obtained. In a previous study comparing the performances of 14 agrammatic aphasics against 45 age-matched controls we obtained the data shown in Table 1 (Crerar, Ellis and Dean ,1996). The tasks were sentence/picture-matching tasks, there were 42 sentences in the test and for each sentence four pictures were offered. The main point to note is the severity of impairment of the aphasic subjects in both in accuracy and speed. The practical effects of this level of impairment include not being able to read for pleasure, not understanding simple written instructions and often finding that text disappears (for example from the TV screen, or as the person travels on a bus) before it can be processed fully. The consequences for the self-confidence, independent mobility and personal safety of such individuals is clear.

Cumulative response latencies (minutes)

Mean no. errors (n=42)

Mean % correct

Control Group (n=45)

Aphasic Group (n=14)















Table 1. Comparison of normal and aphasic performance on a computer-based Microworld Syntax Screening test.

In the study previously referred to we used a computer-based environment called 'Microworld for Aphasia' both to assess and to treat asyntactic comprehension. The software was written in Borland Turbo Pascal® for DOS to run on IBM compatible PCs. To guarantee the reversibility of all sentences created, a miniature artificial world was used comprising only three items, ball, box and star. These could be depicted as either animate characters (stick figures with appropriately shaped heads) or inanimate objects. A similarly limited set of verbs (paints, holds, gives) and prepositions (above, between, beside) were chosen for treatment and a limited set of four sentence structures were treated. The following are typical microworld sentences: The ball paints a box; The ball gives a box to the star; The star is under the box; The ball is behind the star that is under the box.

Using these means, together with a suitable experimental design, we were able to probe agrammatic comprehension in more detail than previous manual methods had allowed. Furthermore, we demonstrated some significant and durable effects of treatment in subjects who were many years post onset, these included very useful improvements in speed and/or accuracy for some individuals (Crerar et al, 1996; Crerar, 1997). The quantitative experimental data was supplemented by subjective reports from patients and carers obtained by questionnaire. These reports indicated much wider psychological benefits of the treatment programme. We found that being in greater control (using the software themselves) and experiencing therapy more as a partnership of equals boosted the self-confidence of individuals and that this manifested itself in other areas of their lives (Crerar and Ellis, 1995; Crerar, 1997).

It became clear during the Microworld study that some of the patients were keen to practice the exercises outwith their allocated appointment time, and that others would have been willing to buy a computer to do so at home. Unfortunately, the Microworld software had been designed as a research tool not a finished product, the input of a trained clinician was assumed, so the software was not suitable to be used unaided. Neither was the software suitable for widespread clinical use, since extending the sentences structures, the vocabulary and the associated graphics could only be done by a computer programmer. We therefore considered the design of a new system to address both these needs.

| Top |

3. Multimedia Microworld

The new software, known as 'Multimedia Microworld' (Beveridge, 1997), was written in Microsoft Visual Basic™ to run in Windows™ on IBM compatible PCs. The main requirements were identified as follows:

  1. Clinicians should be able to create new sentence sets as required;
  2. Pictures should be generated automatically for new sentences;
  3. Patients' performances should be summarised using automatic report generation facilities;
  4. The therapy programs should be usable by patients without trained clinical help.

The first requirement was met by designing an interactive interface providing a set of sentence manipulation facilities (opening previous sentence sets, creation, insertion, deletion, order randomisation, saving). Extending sentences structures and vocabulary needs further work. The second requirement was met by a novel architecture heavily based on linguistic theory, in which the graphics are generated from the deep structure of the target sentence. This has two important implications, one concerns resources, in that no graphics are explicitly stored and the second is that for home based use, only text files containing sentences have to be exchanged. Requirement 3 was met using the integral report generation facilities of Visual Basic, these provide for charts and performance graphs as well as textual output describing a patient's history of interaction. To address the final requirement we incorporated speech output in the system, so that during therapy, sentences are read out, using prosodic stress to focus the particular segment to attend to, e.g., The ball is painting the box. The therapy protocol is rigid; patients work through the tasks, building pictures to match target sentences as the software dictates and when a task is finished, the computer builds the correct picture beside the patient's and declares his attempt right or wrong. Only verb processing has been tackled so far in the new system. Examples of Multimedia Microworld screens can be viewed here.

A pilot evaluation has been carried out with three patients, some key findings are as follows:

  • All three subjects achieved statistically significant improvements on the Multimedia Verb test, from baseline to post-therapy;
  • There was no statistical difference for any of the clients between treated and untreated items, so treatment effects generalised within the Microworld environment;
  • The group achieved statistically significant improvements on the comprehension of Philadelphia Comprehension Battery (PCB)(Saffran et al., 1988) written sentences.
  • The group achieved statistically significant improvements on the comprehension of PCB spoken sentences (though the improvement was less than for the written modality).

The pilot study was conducted by the doctoral computing student who wrote the software (Beveridge, forthcoming). His interaction with patients was confined to fixed formulaic responses (an automation of the mapping therapy protocol of Saffran et al. (1988)). So the results look promising in terms of having provided a therapeutically useful computer system suitable for home use, either by patients unaided, or with the assistance of a carer. It is worth noting that our results were obtained with only 8 hours of treatment, compared with the 60-90 minutes, three times per week for up to 4 months, reported by Schwartz et al. (1994) for manual administration of the same protocol.

| Top |

4. Technology and therapy: new models of service provision

Language therapy has always been a face-to-face activity and usually between a clinician and an individual client. This one-to-one interaction and the personal relationship it engenders are rightly valued. However, one-to-one consultation is expensive and self-limiting, hence the very small amount of treatment that can be offered to any individual patient. There is an analogy in higher education where, although the personal tutorial might be the preferred way to learn, economically such costly tuition is impracticable. In the UK, two principles are emerging in the aftermath of the National Committee of Inquiry into Higher Education (Dearing, 1997): one is the widening of access to individuals who would previously have been excluded and the second is the concept of 'lifelong learning' which envisions individuals learning intermittently and throughout life both for qualifications and also for pleasure. These ideas perhaps have relevance to clinical provision. Just as it seems wrong that access to formal education should cease after a statutory entitlement has expired, so too it seems unduly harsh that language therapy after stroke should be so short. It is clear from our work and the work of others that stroke victims have the capacity to improve even many years post onset. But 'lifelong therapy', that is taking therapy of different forms at different times to suit the client, would not be possible on a one to one basis.

Much as Higher Education is exploring different modes of delivery, in particular flexible and distance learning, so clinical practice might consider something similar. Of course, not all impairments will be equally suitable for remote treatment, but there is clearly a great deal of potential in discovering how we could extend treatment opportunities to those who could benefit, are motivated and yet currently receive no therapy at all. Our work suggests that asynchronous communication between client and clinician is well worth exploring. Asynchronous communication means that client and clinician a) do not need to be co-present and b) do not need to devote the same time slot to their interaction. Thus with Multimedia Microworld, exercises could sent by a clinician as an attachment to an e-mail message and the patient's progress reports could be exchanged the same way. Imagine how effectively a clinician could support a larger case load this way, and further imagine how a 'virtual clinician' (a smart computer program, with a persona) might extend the service further; just as teaching assistants help me to service more students than I could do alone. Electronic communication networks could also be exploited much more to enable patients to support each other, despite their physical immobility or geographical distance.

| Top |

5. Conclusion

We have reported the design and preliminary evaluation of a multimedia system for treating asyntactic comprehension. We have shown that significant treatment effects can be obtained without the involvement of a qualified clinician during therapy sessions. Clinicians can create new exercises with minimal effort and these text files can be transmitted on diskette or via the Internet to remote clients. Telecommunications technology has the potential to transform clinical service provision in the same way that it is transforming Higher Education. Embracing the concept of remote therapy would undoubtedly shift the rôles of associated clinicians more towards needs assessment, monitoring and resource management, but the alternative, 'lifelong neglect', is becoming harder to justify.

| Top |


Beveridge, M. A. (forthcoming). Multimedia Microworld for the treatment of asyntactic comprehension. Unpublished PhD dissertation. Napier University, Edinburgh.

Beveridge, M. A. (1997). Microworld for Aphasia WWW site can be found here.

Crerar, M. A. (1997) The computer as clinical tool: insights from the 'Microworld for Aphasia' study. Invited paper, 5th Symposium on Logopedics, November 1997, Madrid, to be published in Spanish.
English text of paper can be viewed here.

Crerar, M. A., & Ellis, A. W. 1995. Computer-based therapy: towards second generation clinical tools. Chapter in Code, C., & Muller, D (Eds), Aphasia Therapy (2nd Ed). London: Cole & Whurr.

Crerar, M. A., Ellis, A. W., & Dean, E. C. 1996. Remediation of sentence processing deficits in aphasia using a computer-based microworld. Brain and Language, 52/1, 229-275.

Dearing, R. 1997. The recommendations of the 'Dearing Committee' may be viewed here.

Saffran E.M., Schwartz M. F., Linebarger M.C., Martin N., & Bochetto P. 1988. Philadelphia Comprehension Battery. Unpublished.

Schwartz, M.F., Saffran, E.M., Fink, R.B., Myers, J.L. and Martin, N. (1994). Mapping therapy: a treatment programme for agrammatism. Aphasiology, 8, 19-54.

Acknowledgement: We are grateful to colleagues Elizabeth Dean and Alison MacDonald, Queen Margaret College, Edinburgh, who made patients available for the Multimedia Microworld evaluation and to Napier University who funded this work.

| Top | | TIDE 98 Papers |