AAC – Lightwriter, GoTalk and iPad dominate UK sales?

I’m going to use this post to examine an issue of interest to technical researchers in AAC. I’m then going to show how the Domesday Dataset [Red13] can provide evidence to support, or refute, assumptions, uncover important research problems, and map the technological distinctiveness of a user community. This is going to be vastly more `science’ than most of my posts so far, but do hang in there. 🙂

The Domesday Dataset

Note, much of the information in this section is a repeat of the description here. It’s included here for completeness.

In 2013 the Domesday Dataset was created to aid formation of AAC policy at the national level. The dataset records purchases of AAC technology by the UK’s National Health Service between 2006 and 2012, giving information on make, model, price, year of purchase, and geographic area of purchase for each item. It was formed by submitting freedom of information requests to every NHS trust in the UK asking for details of all communication devices provided since 2006. The requests required the year of purchase, and the make and manufacture of each device. The full details of the construction are reported in [Red13].

At the time of writing, the Domesday Dataset contained details of 9,157 purchases from NHS Trusts. [Red13] estimates that the trusts that have responded cover approximately 90% of the UK population. All versions of the dataset are held online and licensed under an Open Data Commons Attribution License. The dataset meets the requirements for three star linked open data according to [BL10]. A sample of information appearing in the Domesday Dataset is given in Table 1. The dataset was not only intended to shape UK policy and research, but also as a snapshot for international researchers: allowing comparison of manufacturers, types of aids, budgets, and prevalence within a tight geographical
domain.

There are, of course, caveats to consider before using the Domesday Dataset. Firstly, for privacy reasons, it is presented with no connection to any other element of AAC provision – it is impossible to match any piece of equipment with a particular user, or to a maintenance record.

Secondly, the NHS does not have the complete information: information from AAC manufacturers show that onlys 44% of sales and 38% of the spend were by the NHS. Even with complete data from the public bodies, researchers would be forced to extrapolate the information, perhaps confirming the trends by means of another research methodology. This work makes the assumption that the relative frequency of AAC purchases and trends in the UK are reflected in the dataset. I am careful not to over-analyse this information, but I do note that having a complete list of NHS purchases, even if they only cover 44% of a county’s purchases, is vastly more detailed than any previous record of AAC provision. Potential problems with the dataset underrepresenting tablet sales are discussed later.









Purchase year Manufacturer Model Num. Unit Price Total Price







2006 Liberator E-Tran Frame 1 £120.00 £120.00
2006 Servox Digital Electronic Larynx 2 £ 520.00 £ 1,040.00
2006 Ablenet Armstrong Mount 1 £190.00 £190.00
2006 Ablenet Big Mack 6 £84.00 £504.00
2007 Inclusive Switchit “Bob the Builder” 1 £49.00 £49.00
2007 Cricksoft Crick USB Switch Box 2 £ 99.00 £ 198.00
2007 Sensory Software Joycable2 1 £49.00 £49.00
2007 Dynavox Boardmaker 1 £209.00 £209.00
2007 ELO LCD Touch Monitor 1 £419.00 £419.00
2008 Ablenet iTalk2 Communication Aid 2 £95.00 £190.00
2008 Attainment Company Inc Go Talk(unknown type) 4 £130.00 £520.00
2008 Aug. Communication Inc. Talking Photo Album 2 £18.91 £37.82







 

Table 1: Extract from the Domesday Dataset, taken from [Red13] (Geograpic
information held seperately)

This post examines the issue that little is known about the prevalence of equipment within the AACuser community, and because of this lack of information it is difficult to establish baselines, or contexts. Perhaps worse, when researchers propose solutions, they must also make a range of assumptions about the applicability of their work to the wider AAC audience. We can, for example, imagine an innovative new model for AAC not being successful because it requires a consistent internet connection, which only, perhaps, 5% of users have. The majority of AAC research is devoted to building up a library of case studies to show the benefits of AAC for user groups. This focus on social issues in AAC research is laudable, and vital for the overall area; however, researchers working in the assistive technology field would be more effective if they could answer direct questions about need, capability and technology. For example, a researcher who must choose between supporting a project that reduces errors in word-prediction for adults using eye-gaze by 20%, or a project that makes Step-By-Step devices more responsive and intuitive to use for children, faces a difficult choice without evidence. If the researcher could check that in a particular geographic area there were 45 eye-gaze systems and nearly 600 Step-by-Steps, then that might influence the decision (at a higher level, this is, of course, the calculation that one expects funding bodies to make when awarding the grants that allow projects to even begin).

Even within the United States, which is the major market for manufacturers, and the most active area for AAC research, the complexities of its healthcare system, differing state legislation, and disability culture make estimation difficult. Even the strong efforts that have been made [MMLBL85, BJ90, BL06, Hue91] give estimations of need and use, but none that can be expected to give the granularity that technologists need for their investigation, or even to frame research questions.

What Domesday tells us

To illustrate the use of the Domesday Dataset for technical researchers, we give some simple results regarding the popularity of various types of AAC device.

Table 2 shows the list of most common ‘high tech’ AAC purchases by the NHS in Scotland, ordered by the number of units purchased between 2006 and 2012. Table 3 gives the same table for purchases in England. Both tables are based on a relatively open definition of ‘high tech’ AAC: these lists include only devices that can produce a range of different utterances, and allow those utterances to be selected by icon, or keyboard. As a result they do not include such devices as, for example: Big Macks; Digital Electronic Larynxs;

Jelly Bean twists; Step–by-Steps; MegaBees and many others, which are included in the Domesday Dataset. As discussed, I do not advise the direct quoting of these figures without first being familiar with the caveats discussed in [Red13]. The figures should be considered comparative only.

Some of the more counter-intuitive results from Tables 2 and 3 include the general absence (with the notable exception of the iPad/iPod) of touch screen devices. Indeed, both the Lightwriter and the GoTalk range comfortably sell more than twice as many units as their nearest touchscreen rival.





Rank Model Units



1 Lightwriter (SL35/SL40) 37
2 GoTalk (all types) 34
3 iPads and iPods 15
4 Springboard Lite 12
5 Vantage Lite 6
6 SuperTalker 6
7 Dynamo 6
8 V Max 5
9 Tech/Speak 32 x 6 4
10 Liberator 14 4
11 C12 + CEYE 4



 

Table 2: The 11 most common ‘high tech’ speech aids purchased by the NHS
in Scotland 2005-2011


A more sobering result to consider for researchers in technical AAC is the popularity of devices that are less obvious targets for customisation and improvement. The GoTalk and Tech/Speak ranges are solid favourites for a particular section of the market and part of their appeal is that they are relatively ’non-technical’and are much easier for users and staff to get to grips with: this appeal is somewhat in tension with advanced features like automatic generation of content and voice banking. It is entirely possible that technical research would have more impact if it focuses on making high-capability devices more acceptable to existing users rather than increasing the already impressive capability of existing
devices.

Another aspect of interest is the speed at which the AAC market changes with respect to the existing landscape: the Dynavox Dynamo, for example, is a popular device in both tables; however it has been discontinued for some time. Finally I consider that there are some systems that I would have expected to appear in these lists that are absent: for example, Dynavox’s Xpress and Maestro or Tobii’s MyTobii, and Liberator’s Nova. Speculating on why some products become more or less popular within this sector is beyond the scope of this work; however, I do consider it an area for future interest.





Rank Model Units



1 Lightwriter (SL35/SL40) 77
2 GoTalk (all types) 74
3 iPad/iPod/iPhone 29
4 Springboard Lite 27
5 V Max 11
6 Dynamo 10
7 SuperTalker 7
8 Vantage Lite 6
9 Tech/Speak 32 x 6 6
10 Chatbox 5
11 C12 + CEYE 4



 

Table 3: The 11 most common ‘high tech’ speech aids purchased by the NHS
in England 2005-2011


This post has shown that examining the Domesday Dataset at even the most basic level identifies a range of factors that can help contextualise the technical landscape for researchers in AAC. To return to the examples given in the introduction, we can see how it would be simple for [SACCB12], to use the iPod and iPad’s position in the marketplace as evidence for the potential of their work and we can see how corpus based approaches such as [MS12] can use the range of AAC devices with internet connections to inform the design
process.

Discussion

Research in AAC policy and technology suffers greatly from a lack of large scale quantitative evidence on the prevalence of devices, and the demographics of users. This work has shown that the Domesday Dataset can be used at the research level to provide context for researchers and to help validate, or not, assumptions about everyday AAC use.

This work also gave an analysis of the impact of the explosion in tablet computing on the AAC technological landscape. It provided evidence that Apple devices are already a significant part of the AAC community and that I expect their presence to grow as older devices are phased out of the market.

At the more fundamental level I hope that this work encourages public debate about where the trade-offs lie in terms of targeting technical research in both AAC and the wider intellectual disability field. It is the author’s position that stakeholders at all levels in AAC should be involved in debate on the areas of focus for research resources.

References

[AC09] Morgen Alwell and Brian Cobb. Social and communicative
interventions and transition outcomes for youth with disabilities
a systematic review. Career Development for Exceptional
Individuals, 32(2):94–107, 2009.


[BBA+07]
Laura J Ball, David R Beukelman, Elizabeth Anderson,
Denise V Bilyeu, Julie Robertson, and Gary L Pattee. Duration
of aac technology use by persons with als. Journal of Medical
Speech Language Pathology, 15(4):371, 2007.


[BJ90]
Karen Bloomberg and Hilary Johnson. A statewide
demographic survey of people with severe communication
impairments. Augmentative and Alternative Communication,
6(1):50–60, 1990.


[BL06]
Cathy Binger and Janice Light. Demographics of preschoolers
who require aac. Language, Speech, and Hearing Services in
Schools, 37(3):200, 2006.


[BL10]
Tim Berners-Lee. Linked data. Personal website
(‘http://www.w3.org/DesignIssues/LinkedData.html’), Jun 2010.


[CKRW11]
L. Coles-Kemp, J. Reddington, and P.A.H. Williams. Looking
at clouds from both sides: The advantages and disadvantages of
placing personal narratives in the cloud. Information Security
Technical Report, 16(3):115–122, 2011.


[DMC+07]
Frank Deruyter, David McNaughton, Kevin
Caves, Diane Nelson Bryen, and Michael B Williams. Enhancing
aac connections with the world. Augmentative and Alternative
Communication, 23(3):258–270, 2007.


[HJ11]
Jeff Higginbotham and Steve Jacobs. The future of the
android operating system for augmentative and alternative
communication. Perspectives on Augmentative and Alternative
Communication, 20(2):52–56, 2011.


[Hue91]
Mary Blake Huer. University students using augmentative
and alternative communication in the usa: A demographic study.
Augmentative and Alternative Communication, 7(4):231–239,
1991.


[HYB04]
EK Hanson, KM Yorkston, and DR Beukelman. Speech
supplementation techniques for dysarthria: a systematic review.
Journal of Medical Speech Language Pathology, 12, 2004.


[LMRH00]
Gregory W Lesher, Bryan J Moulton, Gerard Rinkus, and
D Jeffery Higginbotham. A universal logging format for
augmentative communication. Citeseer, 2000.


[LRMH00]
Gregory W Lesher, Gerard J Rinkus, Bryan J Moulton, and
D Jeffery Higginbotham. Logging and analysis of augmentative
communication. In Proceedings of the RESNA Annual Conference,
2000.


[MMLBL85]
Judy Matas, Pamela
Mathy-Laikko, David Beukelman, and Kelly Legresley. Identifying
the nonspeaking population: A demographic study. Augmentative
and Alternative Communication, 1(1):17–31, 1985.


[MS12]
Margaret Mitchell and Richard Sproat. Discourse-based
modeling for aac. In Proceedings of the Third Workshop on Speech
and Language Processing for Assistive Technologies, pages 9–18,
Montréal, Canada, June 2012. Association for Computational
Linguistics.


[PGM03]
Lindsay Pennington, Juliet Goldbart, and Julie Marshall.
Speech and language therapy to improve the communication skills
of children with cerebral palsy. Cochrane Database of Systematic
Reviews, 3, 2003.


[PGM04]
Lindsay Pennington, Juliet Goldbart, and Julie Marshall.
Interaction training for conversational partners of children with
cerebral palsy: a systematic review. International Journal of
Language & Communication Disorders, 39(2):151–170, 2004.


[RCK11]
J. Reddington and L. Coles-Kemp. Trap hunting: Finding
personal data management issues in next generation aac devices.
Proceedings of the second workshop on speech and language
processing for assistive technologies, pages 32–42, 2011.


[Red13]
Joseph Reddington. The domesday dataset: Linked and open
data in disability studies. Journal of Intellectual Disabilities,
17(2):107–121, 2013.


[SACCB12]
Eva Szekely, Zeeshan Ahmed, Joao P. Cabral, and Julie
Carson-Berndsen. Winktalk: a demonstration of a multimodal
speech synthesis platform linking facial expressions to expressive
synthetic voices. In Proceedings of the Third Workshop on Speech
and Language Processing for Assistive Technologies, pages 5–8,
Montréal, Canada, June 2012. Association for Computational
Linguistics.


[SCRS09]
Tracy A Shepherd, Kent A Campbell, Anne Marie Renzoni,
and Nahum Sloan. Reliability of speech generating devices: A
5-year review. Augmentative and Alternative Communication,
25(3):145–153, 2009.


[Web11]
Tim Weber. Bbc news – google to buy motorola mobility, last
retrieved September 2011, 2011.

Leave a Reply