LEESLAB: Computer-controlled gaze control in CVI-children
Hugo Maes, Erik Vandenbussche, Pat Delaere, Sonja De Lauwe
Subjects: target group
Children with the Cerebral Visual Impairment syndrome (CVI) have visual perception dysfunctions based on congenital cerebral (brain) lesions, not on ophtalmological deficits. Their visual field and acuity can even be near-normal.
The syndrome also often includes motor difficulties ('multi-handicaped').
These children fail to see complex scenes and objects. Objects are often recognized by partial and small details. This is probably linked to their inability to guide their visual attention and gaze direction effectively over the relevant visual field, a defect which constitutes a form of gaze apraxia.
CVI-children do cope relatively well with their visual environment, up to the moment they have to start reading (and writing). They confuse letters; they have problems with word synthesis and with gaze control over the letters in a word, over the words in a line, and over the lines themselves (up down). They fail to adopt the correct left-to-right reading direction and the grapheme-phoneme link is defective. Writing also poses serious problems that are partly due to the motor deficit.
Conventional Low Vision Aids don't work as expected.
Our laboratory is involved in research of CVI, and has developed some effective diagnostic tools in this field (the L94 test). We currently started development of a set of computer programs, specifically intended to address the initial technical reading deficit of CVI children with methods based on gaze control, and the on strengthening the grapheme-phoneme link. It is expected that a much wider group of children with developmental dyslexia and even analphabet adults might benefit from the use of these programs.
The package is called LEESLAB (READLAB).
Material
Pictures Persons, objects, concepts, scenes (2,800)
Textual Text on screen (6,000 words)
Speech + Generated speech (Lernout & Hauspie TTS)
+ Human speech (3,500 spoken words)
+ Spoken phonemes (70).
Lessons Each one containing at least 10 exercises with any of this material: 1500.
Language dependent material
is in Dutch but the programs are open to any other language, provided:

+ another TTS language module is used,
+ the phoneme definition and spoken files are replaced
+ the human speech files (*.wav) are replaced
+ the text material and lessons are replaced.
Methods
The package is called LEESLAB (READLAB), and currently contains
two programs.
+ WOORDLAB (WORDLAB) is mainly word-oriented and interactive,
+ LEZER (READER) is mainly sentence and continuous-text oriented
and less interactive.
In order to address the specific problems of CVI-children, these programs have a number of common features.

Poor Screen		 Many colorful and dynamical screen elements used in popular first grade training software, prove to be attractive to normal children (cartoons, flashing numbers, backgrounds, etc…). Its presence does, however, seriously distract CVI children. Use of such elements is avoided, or they are presented as static and inconspicuous as possible, leaving only the bare essentials on the computer screen.

Single key response		 A majority of CVI children has motor handicaps in a varying degree, which prevent them of easily using a keyboard and/or a normal computer mouse. The program interface has therefore been designed for single key (spacebar, mouse-key) operation. In case of word recognition, it is sufficient to make a single key response when the right word is highlighted amidst a number of distracters.

Grapheme-phoneme strengthening		 Whenever a word is presented graphically, either as text or as a picture, it can be spoken too. Speech is continuous (normal), or by consecutive phonemes.

Gaze control		 Gaze is controlled over text material only, and both in READER and in WORDLAB. The text is presented in the font type, the font size, and the font and background color and contrast suitable for the individual child. To facilitate word synthesis, different groups in the word can be differently colored (first letter, last letter, particular sound groups, or letter
groups).
Apart from these static cues, dynamical cues may be added to assist reading direction and tracking. These consist of left-to-right streaming blocks, arrows or sliders over the word. Consecutive individual letters and phonemes may be highlighted and spoken in the correct reading direction also.
Gaze control is on 3 levels:
Letters within a word (both READER and WORDLAB).
Over words in a phrase or line: words are highlighted in the correct reading direction and may bespoken at the same time (READER).
Flickering blocks show where to start on the next line downward (READER).
Specifics about READER
READER (LEZER) is a computer program that shows a word, a sentence, or several lines on the screen, and reads and/or spells it. Words and/or phonemes are highlighted in the correct reading order. [Nevertheless, words can be pointed at wish for reading or spelling].
Reading speed is computer-imposed, or on the pace of the child. Words may be highlighted and read, or the child may be allowed to read it first, then verify its correct pronunciation.
Pictures matching the scene described may also be shown.
We also are successful using READER to let a CVI child read texts that meet his intellectual development level, but are way above his technical reading skills.
Specifics about WORDLAB
WORDLAB (WOORDLAB) is primarily word oriented, and a child has to make a response to a stimulus.
Stimuli + Text: presentation can be permanent, fixed-period, or tachistoscopic (single flash).
+ Pictures: objects, persons, concepts, actions, and situations.
+ Spoken words either by TTS or by human speaker.
Response + Multiple choice. The correct word is presented amidst a number of distracters. The child can point the correct response, or just hit any key when the computers scan the alternatives.
+ Typing. For those able to type, the correct response can be typed on the keyboard. Complex letter groups or sounds may be mapped on the function keys.
Phonemes and words can be pronounced as the child types it, and feedback is immediate.
+ Speaking. An observer scores the correctness by + or -.
Specifics about computer based training in general
Computer based training of initial readers is not a miracle remedy but in general, offers a number of advantages over pencil-and-paper approaches.
There is complete control over the spatiotemporal aspect of the stimulus material. Dynamic stimuli can be used.
Stimulus presentation is reliable and repeatable from trial to trial.
Responses and latencies can be measured accurately and reliably.
There is immediate feedback on errors and remediation is instant – facilitating the learning process.
The motivation factor is high for the child – more training can be sustained with less fatigue.
To a certain degree, unattended training is possible. Teachers can handle more children with the same input of effort.
Content and setting of the programs is easily changed. Material and training method can be tailored to the child’s training level, skills and interests.
Logging of material used, responses and progress is complete and flawless.
Conclusion
Over the past year, both programs were used with several CVI children.
Two subjects in particular (Nico and Thomas) were followed extensively and daily. The methods, stimulus content and presentation in these programs has been modified constantly to optimize its effects.
A couple of studies currently are under way measuring the differential effect of several stimulus presentation methods, and of the relevance of READER as a reading prosthesis.