More than Words

Augmentative and Alternative Communication (AAC) refers to the systems and methods used by individuals with little or no speech to communicate.

Students who use AAC are at risk of not developing functional literacy skills:

• Even with average or above average intellectual abilities (Dahlgren & Sandberg, 2001; Foley & Pollatsek, 1999; McNaughton, 1993).
• Only 10% of individuals who require AAC have reading comprehension abilities above a grade 2 level, which is well below the attainment of functional literacy skills (Erickson, 2003).
• Poor literacy outcomes increase the risk of social, academic and vocational disadvantage (Foley, 1993; Koppenhaver, Evans & Yoder, 1991).
• Poor literacy skills also limit students’ abilities to effectively communicate on their AAC devices by restricting them to pre-programmed vocabulary and phrases (Fallon, Light, McNaughton, Drager & Hammer, 2004; Millar, Light & McNaughton, 2004; Blischak, Shah, Lombardino & Chiarella, 2004; Nelson, 1992).

Several factors may contribute to students being at risk of poor literacy outcomes including:

• Severe communication impairments are often associated with poorer literacy outcomes (Foley, 1993; Foley & Pollatsek, 1999).
• Different learning opportunities during preschool years due to the impact of comorbidities (sensory, physical and learning disabilities) contributing to less opportunity to play, explore the environment and engage in early literacy activities (Light & Kelford Smith, 1993; Smith, 2005; Koppenhaver, Foley & Williams, 2009).
• Different learning processes associated with learning language using AAC (McNaughton & Lindsay, 1995).
• Limited access to literacy learning experiences and adaptations to enable engagement in these literacy experiences (Sturm, Spadorcia, et al., 2006).

Current reading instruction, particularly phonics instruction, relies heavily on the ability of students to participate by providing verbal responses:

• Students who use AAC require instructional modifications that enable them to participate effectively in literacy instruction without the need to provide verbal responses.
• These instructional modifications involve changing the input and output methods, for example providing a verbal model and creating a selection task for the student using letters or symbols to represent sounds or words.
• Instructional modifications using AAC have been developed for early reading skills such as learning letter-to sound correspondences, sight words and decoding words and single sentences.
• These modifications have not been effectively incorporated into reading instruction using connected text because using existing AAC systems they are not scalable for the amount of text contained in books.
• When students are ready to start reading connected text in books, they are encouraged to read silently using their inner voice (subvocal speech) due to a lack of practical AAC solutions for this level of instruction.

Although the ultimate goal for all students, including students who use AAC, is to read silently, reading aloud is an important instructional stage for students because:

• Development of subvocal speech (inner voice) is a difficult skill for students with severe communication impairments to acquire (Erickson, Clendon, Cunningham, Spadorcia, Koppenhaver, Sturm & Yoder, 2008).
• Reading aloud is overt and therefore provides opportunity for a more direct approach to assessment, error correction and guiding further instruction.
• Reading aloud is motivating for students because they can overtly demonstrate their reading abilities.

Guided by the above principles, this research explores:

• How can we use AAC design to create reading instruction methodologies to move reading instruction beyond words to reading aloud the connected text in mainstream books?
• How can we do this in a manner that as closely as possible emulates the evidence-based instructional techniques and learning opportunities experienced by students who use natural speech to communicate?

Limitations of Existing AAC Devices for Reading Instruction

High tech AAC systems consisting of voice output combined with dynamically configured grids appear to be the most logical AAC choice for enabling students to read aloud the text in books. However, there are many limitations on the functionality of existing systems for reading connected text which may explain the tendency to revert to silent reading for students who use AAC at this stage of reading instruction.

Existing vocabulary layouts in AAC systems are not optimised for reading aloud:

• Symbol navigation is too slow for reading fluently (too many button presses to access each word).
• Symbol navigation (particularly for new vocabulary) requires too much cognitive load when the learning focus should be on decoding and comprehending the text.
• To check the accuracy of a student’s decoding, the instructor needs to enter the exact word and a semantically appropriate symbol for the context of the text into the AAC device – this is very time consuming and not scalable as the amount of text in books increases.
• There is also the dilemma of where to put new vocabulary – if it is scattered it can be used for multiple books but there is a lot of symbol navigation for the student to remember; if words are placed on a topic page although this reduces navigation, access may still be difficult if there is more than one vocabulary page and words have not been ordered according to frequency of occurrence in the book, and, the student will need to learn a new topic page grid location for each word as they occur in new books.
• Books contain different book-specific and often more formal vocabulary than is used when speaking (it is often not desirable to have book vocabulary taking up valuable space in an AAC device if it is not being used frequently in spoken language).
• There is often no easy method to turn the text above symbols on or off so that when reading text above symbols can be turned off to ensure that students are decoding rather than matching text.

The other major problem with existing AAC systems for reading aloud is that there is no capacity to sound out unfamiliar words as they are encountered in the text.

Developing a new AAC Reading System

Based on an analysis of the communication and instructional demands of reading aloud connected text in books, a new AAC Reading System required the following features:

• Vocabulary selection must be quick and easy.
• Vocabulary grid placement must be consistent across books.
• Vocabulary entry must not burden caregivers and educators.
• Students must have capability to sound out unfamiliar words and learn non-decodable words.
• Students must have opportunities to pre-learn symbols for new words and their grid placement prior to reading books.
• The system needed the flexibility to support students to achieve a range of literacy goals from early participation in reading stories to independently decoding book text.
• The system design needed to support reading all types of books including storybooks, decodable readers, levelled readers and plays.

When considering how to speed up vocabulary access and vocabulary entry, it occurred to us that we had one distinct advantage over AAC systems used for spoken conversation – we always know what word should be spoken next because all the words are presented in order in the book text.

The key to the AAC system design was to develop a method of parsing text in the books so that the AAC system always knows the current word, and the meaning of the word in the context of the current sentence in the text.

Parsing the text of the book provides several advantages:

• It enables the base word to be displayed on the grid, reducing the cognitive load of symbol navigation and speeding up symbol selection to maximise reading fluency.
• It enables each word to be correctly sounded out because the semantic meaning of the word in the context of the sentence is known.
• It enables all the vocabulary for each book to be pre-entered and for semantically appropriate symbols for the context of the sentence to be displayed.
• It enables the system to determine when a word in a new book is new vocabulary for the student and to pre-teach the symbol and the symbol position in the grid prior to the student reading the book.
• In plays it enables the system to read the other parts of the play and the student to read aloud their part to practise.
• It enables the development of an error correction process when the student makes an incorrect symbol selection.

Once we had successfully developed a methodology to parse the text in books, the next challenge was how to optimise the symbol layout for the text. The goal was to always display the base symbol on the grid for the current word being read.

For reading, a completely new way of displaying vocabulary was required.

Through prototyping and repeated testing with book data we determined that the sheet layer model was the best method of storing and displaying vocabulary for reading:

• Each word is assigned to a fixed position in a cell within a grid layer based on word type when it is entered as a new word into the dictionary (i.e. when the book is parsed).
• The position of every word remains constant regardless of the book being read.

The cell position assigned to a word is determined by three factors:

• The Fitzgerald classification of the word (Fitzgerald, 1929).
• The semantic relationship of the word to other words.
• The similarity of the symbol to existing symbols on the grid (similar symbols occupying different cells in close proximity is undesirable).

In the AAC Reading System, the Fitzgerald classification of words is represented by a series of layers which includes a verb layer, noun layer, adjective and adverb layer, pronouns and people layer and a miscellaneous layer for all the words that do not fit within the other layers.

The following process is used to assign words to cells:

• Each word is assigned to a specific sheet layer based on its word class (e.g. noun, verb, adjective etc.).
• The cell position that the new word will occupy within the layer is assigned by the relationship between the new word and existing words within the dictionary.
• Words are positioned in cells (which correspond to display positions) within sheets and across layers so that words that are related in meaning but have different word classes (Fitzgerald types) occupy the same cell on multiple layers.

For example, the words 'use', 'user' and 'useful' are all semantically related words but have different word classes so the words appear on different layers but occupy the same cell position. The sheets and layers displayed dynamically change on a word-by-word basis as the student reads the text of the book.

Grouping related words in the same cell position and in their grammatically correct layers reduces learning load by ensuring similar words appear in a single predictable location. The sheet and layer data structure present every word in the English language (and most other languages) in a way that is accessible with one or two button presses, with most words accessible with one button press. Combined with the consistency of the location for each word this substantially improves reading fluency.

This system also makes it possible to display a symbol that is semantically congruent with the meaning of the word in the context of the sentence, for example the different meanings of words in different word classes such as the noun ‘spy’ (the person) and the verb ‘spy’ (the act of watching another person’s actions).

Provision of semantically congruent symbols enhances reading fluency by providing symbols that make sense in the context of the sentence which reduces the symbol search time. It also supports reading comprehension and vocabulary expansion by providing an accurate depiction of the meaning of the word in the specific context of the current sentence.

We are currently working on programming the system to have an overlay layer displaying symbols adjacent to the target symbol, that contain similar graphemes to the target word. So for example, next to ‘cat’ you might also have ‘cot’, ‘bat’ and ‘cab’. This will reduce the temptation to only decode the first grapheme in a word and then guess. It will also provide information about error patterns for the student which may also guide further instruction.

Decoding and AAC

There has been considerable concern about how individuals who use AAC can develop phonological awareness skills, phonological recoding of written words and phonological coding in short term memory because these skill areas were thought to be reliant on the ability to articulate sounds either overtly or internally through subvocalization (Blischak, 1994; Foley and Pollatsek, 1999; Millar, Light & McNaughton, 2004).

To discuss why a congenital absence of speech may be problematic requires a brief review of short-term working memory. Baddeley (1997) proposed a model of working memory consisting of a controlling central executive system and two slave systems: the visuo-spatial sketchpad and the phonological loop.

The visuo-spatial sketchpad is used to manipulate visually presented information.

The phonological loop:

• Assists with the ability to manipulate speech-based information.
• Consists of a phonological store where speech-based information can be retained for approximately 1.5-2 seconds and an articulatory control process.
• Articulatory control process uses subvocalization to perform two tasks:
  • Rehearsal to maintain and refresh information contained in the phonological store.
  • Converting written material such as text into a phonological code that can then be sent to a phonological store.

Reading successfully involves acquiring the ability to translate the printed text which is perceived in the visuo-spatial sketchpad into phonological code which is produced and maintained either verbally or through subvocal speech through the articulatory control process to derive it’s meaning (Erickson et al, 2008).

This process is particularly important in the early stages of reading when students are thought to be more developmentally dependent on phonological processing to support decoding (Stanovich, 1986).

It was originally hypothesized that the ability to perform subvocal phonological coding may develop initially through the use of overt natural speech (Card & Dodd, 2006)

There have been mixed research findings regarding the ability of students with severe communication impairments to acquire skills thought to rely on phonological coding (Bishop and Robson, 1989; Foley & Pollatsek, 1999; Vandervelden & Seigel, 1999). Some differences in research findings may be attributable to variations in participant characteristics, experimental tasks and the methods of analysis (Card & Dodd, 2006).

The current consensus is that using natural speech is not essential to enable the acquisition of decoding skills, but the methodology of how students acquire decoding skills and the rate at which they acquire these skills when they can’t use natural speech may be different (Sturm, Spadorcia, et al., 2006).

Students who use AAC are likely to have more difficulty acquiring decoding skills and are likely to be less efficient at using them than people with natural speech (Foley & Pollatsek, 1999).

It is not known whether the relationship between poor phonological coding and decoding skills is causal or reflects the cyclic interplay between these factors where difficulty performing skills leads to less practice of skills creating a Matthew effect (Foley & Pollatsek, 1999; Stanovich, 1986). The Matthew effect is where students who enjoy reading read more and become more proficient whilst students who dislike reading read less and have a slower rate of improvement and over time the gap between the two types of students widens (Stanovich, 1986).

Instructional Design Modifications for AAC

Phonics based instructional approaches are effective methods of teaching literacy skills, particularly for students at risk of, or with identified reading difficulties (Adams, 1990; Carnine, Silbert, Kameénui, Tarver & Jungjohann, 2006; National Institute of Child Health and Development, 2000).

Students who use AAC benefit from the same explicit systematic phonics instruction as their peers. To successfully participate in phonics instruction, students who use AAC require instructional adaptations that compensate for their inability to use natural speech to:

• Blend and segment sounds
• Identify letter to sound correspondences
• Read words, sentences and passages of connected text

Current research has focussed on developing alternative AAC methodologies for early reading instruction with an emphasis on:

• Sight word recognition
• Developing letter to sound correspondences
• Decoding single words and simple sentences.

The vast majority of research into phonics instruction for students who use AAC has involved modifying the input and output methods of tasks to compensate for the inability of students to provide a verbal response (e.g. Blischak et al, 2004; Banajee, 2007; Benedek Wood, 2010; Fallon et al., 2004; Harwood, 1996; Johnston, Buchanan et al, 2009; Johnston, Davenport et al, 2009; Millar et al, 2004; Truxler & O’Keefe, 2007).

Examples of adaptations included:

• Change the task demands by having an instructor provide a verbal model and then the student providing a response by identifying a corresponding word, picture or letter.
• When working on blending sounds to sound out words an instructor verbally sounded out the word and the student selected a corresponding picture, word, or symbol from an array.
• In letter-to-sound correspondence tasks, adaptation involved an instructor saying the sound and then presenting the student with a letter array.
• In word recognition tasks, adaptation involved presenting a written word and then presenting the student with a picture array, or, presenting the student with a spoken word and then requiring them to identify the printed word from an array.

Another important instructional priority for students who use AAC is supporting their ability to develop and use inner speech during reading. One example of an approach that specifically targets this is the nonverbal reading approach.

The nonverbal reading approach:

• Uses systematic direct instruction to explicitly teach students to develop the metacognitive ability to use internal speech (Heller, 2002; Coleman-Martin et al, 2005; Swinehart-Jones et al., 2005).
• When teaching single word recognition tasks, three specific steps are taught (Coleman-Martin et al., 2005):
  • Students are instructed to say the individual sounds in words using internal speech with an oral model from a teacher/computer.
  • Students learn to blend the sounds together using internal speech whilst being provided with an oral model by the teacher/computer.
  • Students are instructed to say the sounds fast using internal speech while the teacher/computer says the word.

In the AAC Reading System we extend on these methodologies and apply them to the decoding of connected text in books. Students are presented with the text of the sentence in the book with the current word highlighted and are provided with a grid array of symbols containing the base word for the current symbol. If the student is unable to independently decode the word, they can select the sounding out button to sound out the word.

The sounding out process follows the steps described above in the nonverbal reading approach, and it is important that students are instructed to use their inner voice whilst completing the task.

Adaptation of instructional tasks to make them accessible for students who use AAC inevitably alters both the learning task and the associated learning demands (Blischak, 1994). These modifications are necessary, but it is important to understand the implications of these in terms of the change to learning tasks and the learning demands of the specific response methods, and, the underlying impact of using an AAC system to develop language:

• By using alternative methodologies where students are presented with a response array, the learning demands of the task are changed from a recall task to a recognition task (Blischak, 1994).
• Students who require AAC process symbol based graphic representational systems rather than an oral representational system for spoken words – it is not known how this impacts on their ability to acquire the phonological processes necessary to decode words (McNaughton & Lyndsay, 1995).
• When reading, students who require AAC have additional processing demands of translating print into meaning to select the correct symbol whilst their peers with natural speech can initially focus on the simpler translation of print to sound (Erickson et al., 2008).
• A student who uses AAC must have a word in their receptive and expressive AAC vocabulary in order to successfully select the corresponding symbol from a symbol array (Erickson et al, (2010).
• There is the dilemma of what to do with the words that have abstract meaning (e.g. is, the, at) and are therefore not easily represented as symbols – given many of these words are non-decodable anyway - should they be represented as words as they typically are in AAC devices with an acknowledgement that this reduces the task to match to sample for these words, or, should students learn a set of abstract symbols specifically for the purpose of decoding and or recognizing these words as sight words? We are still unsure of the best methodology and plan to provide educators with both options until further research provides a definitive answer to the best method of representing these words.

The unique demands of learning to decode using symbols necessitated the inclusion of a learning system to track the student’s exposure to symbols in previously read books and provide opportunities to pre-learn new symbols and their grid locations prior to reading the text so that their cognitive effort can focus on decoding the text when reading the books. Inevitably a student may be in a position where they are able to decode a word but have forgotten the corresponding symbol. We also provide a symbol learning process within the books so that a student may be reminded of the symbol and the corresponding grid location whilst reading. Without the provision of these systems, it would be difficult to pinpoint whether a student is having difficulty decoding a word or recognising/finding the corresponding symbol.

Providing appropriate supports also reduces the tendency for students to resort to guessing.

Rasinski & Padak (2013, p 13) describe fluency as the ability “to read expressively, meaningfully, in appropriate syntactic units (phrases, clauses), at appropriate rates, and without word recognition difficulty.”

Word automaticity and prosody are considered the two central components of reading fluency. Developing word automaticity frees up cognitive resources to concentrate on the meaning of the passage and prosody helps convey the meaning contained within the specific text passage using appropriate phrasing and expression (Rasinski & Padak, 2013).

When reading, background knowledge, being able to make connections between this knowledge and the content of the text, vocabulary knowledge including the context, meaning derived from morphology and the word meanings; and structural knowledge such syntax and the text genre all combine to support reading comprehension (Sturm, Spadorcia, et al., 2006).

Designing instruction to support reading fluency for students who use AAC is daunting because every component of fluency poses instructional challenges. Research to date has not addresses the challenges of reading fluency for students who use AAC. This may reflect both the inherent difficulties of developing alternative methodologies for reading fluency for these students and the fact that most students are instructed to read silently at this level so there is difficulty obtaining overt measures.

Despite these difficulties, we know developing reading fluency is of critical importance to becoming a proficient reader. Reading fluency is often considered the bridge that supports students to move from focussing on decoding to reading with more automatic word recognition and an emphasis on reading comprehension (Rasinski & Padak, 2013). If we want more students to move beyond the attainment of a grade 2 level of reading comprehension, then developing ways to support the development of reading fluency must be at the forefront of future research into instructional design.

How can improvements in AAC design support the development of word automaticity?

Whilst AAC technology will not address all the issues related to developing reading fluency, we believe that there is potential for technology to contribute to the development of some of the key components of reading fluency. The aim is to maximise the opportunities to participate in instruction and improve as many components of reading fluency as possible.

Greater opportunities to practise & benefit from instruction through reading aloud.

At a general level, providing access to a range of books without the need for any vocabulary entry by caregivers or teachers provides greater opportunities to practise reading and re-reading texts both independently and during reading instruction. Words are not read with automaticity until students have been exposed to them many, many times.

Although students can access books using silent reading alone, the overt nature of reading aloud allows for greater opportunities for correction and praise for accurate reading. There is also the opportunity for students to overtly demonstrate their abilities which is highly motivating for students. When students are actively encouraged to use their inner voice to read along with the synthesised voice, they are also having greater opportunities to work on developing their inner voice for silent reading using a verbal model.

Speed of Word Selection

Speed has always been the enemy of AAC systems. The time taken to compose messages using AAC systems compared with speaking is significantly slower. Fortunately, because of the predictable nature of the book text, it is possible to speed up symbol selection in the AAC Reading System by parsing the book text so that the base symbol for the current word is always displayed in the grid. Once a student recognises or decodes a word, symbol selection in the AAC Reading System is fast by AAC standards, particularly for students accessing symbols through direct selection. Alternative methodologies for obtaining measurements of reading rate may be possible by examining latency times between the current word being displayed and symbol selection by the student. Over time reductions in latency periods may provide an indicator of increasing word recognition automaticity. In the case of students who use AAC, the definition of word recognition automaticity may need to be defined more broadly to encompass both the decoding or recognition of the current word and selection of its associated symbol. When obtaining an equivalent measure to words per minute for students using AAC, three additional factors will need to be considered in the calculations:

• The number of button presses required to read the word varies between words, with the majority of words requiring a single button press, but some words (e.g. words with tense, plural and/or possessive markers) requiring two button presses.
• Recognition of some words may be a case of match to sample (e.g. if words like ‘is’, ‘the’ and ‘a’ are represented as words due to the difficulty of representing them as meaningful symbols).
• Symbols vary in their opaqueness and also in their visual similarity (e.g. pronouns look very similar without a text label) which may delay selection even if a word is decoded correctly quite quickly.

Consistent Grid Location of Symbols

Searching for the corresponding symbol occurs after word recognition has occurred, however having a consistent grid location for every symbol that stays the same every time a word is used means that students will minimise search times for symbols once they have learnt their location. This improves fluency and automaticity by minimising the cognitive energy diverted to locating symbols in the grid. The way the sheet structure within the grid is designed to display related words in the same grid space also means that if a student knows a word e.g. spy (verb) and then encounters spy (noun) in the text, the grid location is consistent and an appropriate symbol is displayed. This improves symbol selection fluency and also provides opportunities for discussion to enhance the student's understanding of how words are related despite having different symbols to reflect their different semantic meaning.

How can improvements in AAC design support the development of prosody?

Prosody – the ability to read with expression, phrasing and meaning – is the most challenging aspect of fluency to target using AAC technology. The challenges and possibilities for prosody instruction are explored with references to Rasinki and Padak’s Multidimensional Fluency Scale (2013, p28; adapted from Zutell and Rasinski, 1991). In this scale students are given a score of 1 (lowest) to 4 (highest) on their ability to execute four dimensions of fluency whilst reading a text passage:

• expression and volume
• phrasing
• smoothness
• pace

We would suggest that the first dimension, expression and volume is very difficult to assess and achieve using current AAC technology, but there is potential to adapt the rating scales for phrasing, smoothness and pace to provide some indicators of a developmental progression in the acquisition of fluency skills for students who use AAC. Some of these areas have already been explored previously with reference to automatic word recognition but they also play an important role in reading with appropriate prosody.

Reading with Expression & Volume

This is one area that is very difficult to address using AAC and may be best to teach using alternative methods such as a verbal model with the student attempting to emulate similar pace and expression as the instructor using their inner voice. This teaching can be made more explicit by jointly examining the text and identifying words or groups of words where intonation should change, words that should be spoken louder and places where there should be pauses. Other examples may include using sentences where the way the words are spoken changes the meaning of the sentence and looking explicitly at how making these changes modifies the meaning.

The ability to read and indeed speak with expression when using AAC is limited by the technical capabilities of using synthesised speech technology. It is possible to:

• indicate rising intonation for questions and exclamations
• boost the loudness of some word to add emphasis to words that are bolded or capitalised in the text
• and to add pauses within the text for punctuation.

However, that is the limit of the technology. The other problem is that all of these elements have to be pre-programmed into the AAC Reading System– there is no way for the student to modify these during reading to highlight their own interpretation of the meaning of the text. Meeting this challenge is unfortunately beyond existing technology capabilities at this stage.

One modification that occurs in the AAC reading system is that when the student has read aloud the sentence, the AAC reading system rereads the sentence at normal pace using as much expression as it is possible to program so that the rate and expression is as natural as is possible with synthesised voices. This also helps to maintain comprehension and recall of the sentence given it can take longer to read using AAC than natural speech.

Phrasing

Phrasing is one area where there is potential for the AAC reading system to support explicit instruction before students make the transition to silent reading. It is not possible to address some of the intonation patterns or the use of stress or pauses within the text for punctuation because they are controlled by the AAC system rather that the student. However, it is possible to examine how the student is using phrasing and to explicitly teach appropriate phrasing by gradually extending the number of words read as a unit (with subsequent symbol selection) until they reflect appropriate clause and sentence units. This could be examined by measuring latency times between symbol selections within clause and sentence units and determining whether the larger pauses occur at clause and sentence boundaries.

Smoothness

Smoothness of reading refers to how often students pause whilst reading and how often they need to sound out words whilst reading. Recognising that the definition of smoothness will need to encompass symbol recognition and navigation as well as word recognition, by examining mean latency periods from words being displayed and symbols being selected and collecting data on the frequency that a student needs to select sounding out to decode a word it may be possible to document improvements in the smoothness of reading over time, particularly when assessing re-reading of books.

Pace

Pace is concerned with the speed of reading and whether the pace is even throughout reading the text. Data collection within the AAC Reading System can assist with documenting reading pace. Also, if a student’s mean latency time from a word being displayed and a symbol selection is known, in the future the AAC Reading System could be used to improve a student’s reading pace by encouraging them to read at a slightly faster pace by matching their symbol selection to a verbal model paced at a speed customised for the student.

A Final Important Note about Fluency and the AAC Reading System

The AAC Reading System has been developed to try and fill an instructional gap by enabling students to read aloud connected text in books whilst they are developing their ability to use their inner voice. This is important in the initial stages of reading because it makes the reading process overt and enables students to demonstrate their knowledge and also benefit from instruction and error correction when errors occur. However, once a student reaches a level of reading proficiency where they have good automatic word recognition, they understand phrasing, and the length and complexity of sentences within the text is increasing, they should transition to using their inner voice as their main reading method as this will be more efficient. Continued reliance on the AAC Reading System beyond this stage is likely to start to slow down their reading fluency rather than enhancing it, and the amount of text is likely to make ongoing symbol selection fatiguing even when symbol navigation has been minimised.

We have explored how AAC technology can be applied to the unique communication and instructional demands of reading to develop a customised AAC Reading System. AAC systems for reading need to include:

• opportunities to sound out words
• opportunities to pre-learn symbols
• preloaded vocabulary
• efficient symbol navigation

Unique issues relating to learning phonics using AAC, learning to read using symbols, and developing reading fluency using AAC need to be considered when designing instruction and assessment of reading skills for students who use AAC.

Many factors contribute to students who require AAC being at risk of literacy difficulties. Difficulty participating in instruction is a factor that can be addressed through both instructional design and design of AAC systems. This research sought to extend the existing literature to examine how AAC systems could be redesigned to support reading aloud the connected text in books. Through redesign it was possible to develop a system that will enable students to read aloud whist minimising the cognitive load associated with symbol navigation.

There is still considerable research needed to explore how to support students to read connected text, and many potential improvements to be made to our AAC Reading System, however we hope to stimulate research and discussion on how AAC design may be adapted to ensure we maximise the ability of students who require AAC to participate in reading instruction.

This is a very short video showing a tiny part of the AAC Reading System functionality. We will be back soon with a more detailed video.

Acknowledgements