JCPSLP Vol 22 No 1 2020

(Graham, 2014) or be used to support or cue spoken language skills (e.g., Nickels, 1992). We now discuss three different types of treatment that have been used with graphemic buffer impairment in particular (for reviews, see Beeson [2004], Krajenbrink, Kohnen & Nickels [2015] and Thiel et al. [2015], and Graham [2014] for a review of treatment options for dysgraphia in progressive aphasia). Lexical treatment A first method of treatment focuses directly on improving the spelling of a specific set of words. Beeson (1999) described two methods for implementing this type of lexical treatment using repeated exposure to and copying of words in an Anagram and Copy Treatment (ACT) and Copy And Recall Treatment (CART). ACT involves rearranging the letters of a word provided in a scrambled order (with or without extra, distractor, letters), followed by copying the word and recalling it from memory. CART treatment involves the participant copying target words that are labelled pictures, followed by a written naming task with those pictures to test recall. For individuals with good reading (and reading comprehension) CART treatment has also been used without a picture present (e.g., Raymer, Cudworth, & Haley, 2003). Several studies have reported improving spelling of treated words using treatments incorporating these principles of copying and recalling spellings (with and without pictures) for individuals with dysgraphia, including those with graphemic buffer impairment (see Krajenbrink et al. [2015] for an overview). How might such a treatment improve word spelling in people with graphemic buffer impairment? Rapp and Kane (2002) suggest that one possibility is that the stored lexical representations (in the orthographic output lexicon) of treated words may become stronger and therefore more resistant to graphemic buffer damage. This results in improvement of only the treated words. However, if treatment also results in improved spelling of words not used in treatment, it is possible that graphemic buffer functioning has been improved with greater ability to keep representations active, or strengthened processes within the buffer, such as increased speed of converting abstract letters into their written form (Panton & Marshall, 2008; Rapp & Kane, 2002). However, this generalised improvement seems to occur less often than improvement on just the treated words. Treatment focused at the level of the graphemic buffer Not many studies focus on directly improving the functioning of the graphemic buffer (Beeson, 2004). However, a number of studies have been described that instead train the participant to use other intact language abilities to circumvent the problems relating to orthographic working memory. The benefit of this approach is that a successfully taught strategy can be used for the spelling of not just the words used in treatment but any word that the participant may wish to write (Krajenbrink et al., 2015). Error detection Some studies have capitalised on a participant’s unimpaired ability to recognise errors, and to use intact phoneme- grapheme correspondences to correct errors and spell unknown words (Hillis & Caramazza, 1987). Segmentation De Partz (1995) focused on compensating for difficulties of a participant with buffer impairment (as well as some

additional spelling impairment) in spelling long words by teaching him to divide long words up into smaller segments. This was a useful strategy for the participant, resulting in fewer errors. Cardell and Chenery (1999) included a phase of phonological segmentation therapy as part of their spelling treatment of a mixed case of dysgraphia (including buffer impairment). During this phase, the participant was taught to segment nonwords into their individual sounds, which resulted in improved spelling of untreated nonwords as well as improved written naming. The authors argue that their treatment not only improved the ability to convert phonemes into graphemes but also enhanced the functioning of the segmentation process within the graphemic buffer. Including a segmentation training could therefore be a promising treatment approach, especially for mixed dysgraphia; however, more studies are needed to investigate whether this can be a helpful treatment to support graphemic buffer impairment specifically. Using writing aids When poor writing skills are accompanied by relatively intact spoken language skills, treatment can make use of those intact skills by using voice recognition software to provide a means to produce written text and compensate for poor writing ability. For example, Bruce, Edmundson and Coleman (2003) described an individual with buffer impairment that resulted in very slow writing. Spoken language was also impaired but was superior to written abilities, and therefore the use of voice recognition software was successful in improving speed and accuracy of producing written output using a compensatory speech- recognition system. Outcomes of treatment When spelling treatment is successful, at the very least the items targeted in treatment should have improved. Even more desirable would be generalisation of the results of treatment where untreated items also improve, and/or treated items are successfully used in a different writing context, ideally in functional, everyday writing contexts. However, so far it has been difficult to achieve this desired outcome of treatment in graphemic buffer impairment. A number of studies have examined whether treatment of the graphemic buffer is likely to result in generalisation (e.g., Rapp & Kane, 2002; Sage & Ellis, 2006). Considering the graphemic buffer is involved in spelling of all words, improvement of the buffer should result in better processing of not only items targeted in treatment, but also of any other words that are spelled. Indeed, Rapp and Kane (2002) found generalised improvement to untreated words for their individual with graphemic buffer impairment, compared to only treated-item specific effects for the participant with lexical impairment. Unfortunately, however, other individuals with buffer impairment have not shown generalisation to untreated items following treatment (e.g., Krajenbrink, Kohnen, & Nickels, 2017). Furthermore, Thiel, Sage and Conroy (2016) found no transfer of treatment effects to the everyday writing context of writing emails. As Thiel et al. note, there is a need to better understand how the gains of treatment can be transferred to everyday writing (see also Pettit & Trope, 2018). Therefore, if generalisation after treatment cannot be guaranteed, it remains essential in clinical practice to select treatment items with functional relevance for the individual (Renvall, Nickels, & Davidson, 2013), so that any improvement on treated items will have maximum benefit.

19

JCPSLP Volume 22, Number 1 2020

www.speechpathologyaustralia.org.au