iPad Apps in Teaching Programs for Kids with Autism ... ? iPad Apps in Teaching Programs for Kids
Post on 17-Jul-2018
iPad Apps in Teaching Programs for Kids with Autism Spectrum
Instituto Superior Tecnico, Lisboa, Portugal
From day to day, the use of technology as a teaching tool has grown. The presentation of educationalexercises through electronic devices reveals itself as more attractive and captivating to the user whencompared with traditional methods. The objective of this thesis was to develop an attractive mobileapplication that would make the development and acquisition of learning skills easier to childrendiagnosed with autism spectrum disorders. With that in mind, we researched about autism, its impacton children and traditional teaching methods available and used. Also, we researched how technologyis used as a teaching tool in order to understand what are the methods and techniques most successfulin an application of this kind. After the research was done, and after we concluded that the possibilityof customization was one of the keys to success, we designed and architect one first version of thisapplication. This version was then tested and submitted to several changes based on the feedback ofa professional specialized in Special Education and Rehabilitation. The last implementation stages ofthis application were submitted to tests with a user diagnosed with an autism spectrum disorder, inorder to test and enhance its efficiency. The results of the evaluation met our objectives given thatthey show that there is in fact a improvement in reading skills.Keywords: educational exercises, electronic devices, mobile applications, children, autism spectrumdisorder
According to statistics in the United States ofAmerica (USA), 1 in every 68 children (from birthto age 8 years) have been diagnosed with an AutismSpectrum Disorder (ASD), and boys are four to fivetimes more likely than girls to have it. 
Although there is no statistics for Portugal, if wetake this numbers into account and consider that inthe last population census (2012) there were 994835children aged from 0 to 9 years  we can make anestimation of 113.600 children being identified withASD.
ASD refers to a group of development brain dis-orders. It is called a spectrum due to its wide rangeof symptoms and levels of impairment or disabil-ity. Most certainly a child suffering from ASD willhave difficulties in social interactions and commu-nication, and will engage in repetitive behaviours.
Recent studies   show that the use of iPadsand other related devices in educational and reha-bilitation problems can help individuals with ASD,from mildly impaired to severely disabled.  Its in-tuitive and uncomplicated way of use are some ofthe aspects that make it so helpful and a good tool
to approach individuals with such characteristics asthe ones with ASD.
2. Background2.1. Autism
The term autism, as it is known today, was firstused in a paper  in 1943, written by Dr. LeoKanner, a child psychiatrist and physician at JohnsHopkins University in Baltimore, USA. He observeda group of eleven children (eight boys and threegirls) with ages below 11, who were considered byothers schizophrenic and feeble-minded, and con-cluded that all of them had difficulties in relatingwith other people, communicating and engaged inrepetitive rituals, but were all unquestionably en-dowed with good cognitive potentialities and thatalthough there are many similarities with childhoodschizophrenia this condition had different particu-larities. 
Today this disorder still continues to be describedby similar symptoms, namely difficulties in socialrelations and interactions, problems with commu-nication and repetitive behaviours and actions.
Since May 2013, with the publication of Diagnos-tic and Statistical Manual, fifth edition, (DSM-V)
all autism disorders were merged into one spectrumdiagnosis of ASD, no longer being divided into dif-ferent subtypes. For example, a children that wouldpreviously be diagnosed with Aspergers syndromewould now be diagnosed with ASD without lan-guage or intellectual impairment. 
ASD is diagnosed by physicians and psycholo-gists, based on behavioural evaluations, but theparents are the ones who usually notice the firstsigns. 
2.1.1 ABA - Applied Behaviour Analysis
This therapy combines decades of research in be-haviour modification. In fact, it goes back to 1900swhen Ivan P. Pavlov found what he called condi-tional reflex and Edward L. Thorndike describedthe law of effect. This law states that when a be-haviour produces a pleasing effect on the surround-ing environment, there is a higher probability of itoccurring again in the future, under similar circum-stances.
This is the main principle behind ABA, whichis called positive reinforcement. For example, if achildren receives a candy for each time she eats thewhole soup, she is more likely to eat the whole soupin the future. To be effective a reinforcement shouldimmediately follow the behaviour.
Reinforcements can be scheduled to be contin-uous or intermittent, on the first option, the be-haviour is reinforced every time it occurs. If theschedule is defined as intermittent, the person willreceive reinforcement while learning or engaging ina new behaviour, which is called acquisition, andonce the new behaviour is acquired, the reinforce-ment will be delivered intermittently, which is calledmaintenance.
There are four more principles in ABA, namelyextinction, punishment, stimulus control and re-spondent conditioning. Extinction, in oppositionto reinforcement, has the objective of weaken a be-haviour, if an action is no longer reinforced it is ex-pected to extinguish. For example, everyday Johngoes to the vending machine at his work and spendsa one Euro coin on a beverage. For two days in arow John spent the coin but the machine did notgive him his beverage. Since the behaviour stopped,it is not being reinforced, he stopped going to themachine and bought his beverage at the coffee store.
Punishment, such as extinction, is supposed to beused to weaken a behaviour. For instance, imaginethat Mary was cooking and whatever she had on theiron pan started burning. Her immediate reflex wasto grab the pan, with bare hands, and to take it outof the heat. However, as soon as she felt the painshe dropped it. After this happened Mary alwaysuses something to grab things that are hot. Herbehaviour had an immediate outcome that was not
desirable and because of that it is less likely thatshe will repeat it in similar conditions.
All these principles are based on behaviour mod-ification and that a behaviour is controlled by itsconsequences. Stimulus control is used by analystswho try to understand what outcome a behaviourhas and how they can manipulate it in order to mod-ify the behaviour. Reinforcement, extinction andpunishment are related to a certain situation, wheresome behaviours can be reinforced under some con-ditions and punished under another. For example,Michael always asked his grandparents for candybut only his grandmother would give it to him. Astime passed, Michael only asked his grandmotherbecause he knew his grandfather would not give himcandy. Basically, his grandmother reinforced his be-haviour, while his grandparent extinguished it. Wecan say this behaviour is under stimulus control be-cause it only happens under a specific stimulus hisgrandmother.
The last of the basic principles of ABA is respon-dent conditioning. This can be exemplified as fol-lows: imagine that Anna takes the same way homeevery night, and in some days, at a certain place,she hears a scary sound that makes her heart beatfaster. Now every night, with or without the scarysound, when Anna gets to that specific place herheart starts racing. What first is a normal reactionof the body (reacting to the sound) becomes a con-ditioned response (reacting to the place where thesound occurs). In other words a form of learn-ing in which a response is elicited by a neutralstimulus which previously had been repeatedly pre-sented in conjunction with the stimulus that origi-nally elicited the response. 
2.1.2 ABA and AutismThere are several approaches to ABA therapy butall of them use similar methods. A typically ABAtherapy is applied twenty-five to forty hours perweek and each session lasts two to three hours, ac-cording to the childs needs and skills. There istime devoted to learn specific tasks (usually be-tween three and five minutes) and at the end ofeach hour, ten to fifteen minutes breaks are taken.These breaks are also used for incidental teaching orpractising skills in new environments. Every ABAmethod uses a three step process that is intendedto help determine why a behaviour occurs. Thisprocess is known as the ABC:
Antecedent A verbal or physical stimulus, likea request or command;
Behaviour A resulting behaviour, response orlack of response to the stimulus;
Consequence Positive reinforcement, for thedesired behaviour or no reaction otherwise.
All of the approaches use a reinforcement whena desired behaviour occurs. This reinforcement hasto be seen by the child as a reward, and becauseof that, it should be continually evaluated if thechild still feels like it is worth to work for it. Thereinforcement should only be available to the childas a reward or else she can lose interest and it willlose its power.
A variety of reinforcements should be used, notonly to keep the child interested but also as a differ-entiating element. If the child has several reinforce-ments, the ones that are their favourite should beused as a reward for the most desired behaviours.The reinforcement should be appropriated to thechilds age and, whenever possible, it should be pre-sented in different ways. Surprises are very well ac-cepted and highly motivational.
Social reinforcements, like smiles and compli-ments, should also be used in association with phys-ical rewards.
Reinforcements should not be used as a bribe,although this works in a short term. In the longterm, it can create serious problems.
ABA is not only used to modify behaviours butalso to teach. The teaching of a new skill is made bybreaking a complex task into smaller steps or com-ponents. This is called task analysis and is used byeveryone on everyday chores, like following a recipe.It is specially useful for children with ASD giventheir learning needs. This can be used for any skilland the number of steps will depend on the childsability.
Prompts are also used when teaching skills. Aprompt is a question or instruction given with theintention of directing the individual to engage a cer-tain response or behaviour. As the child masterseach step, the prompts are gradually eliminated.This process is called fading.
Another thing that is common to the ABA ther-apies is the assessment of the child. Frequently theprofessional responsible for the therapy assesses thechild in order to keep track of the childs learningand efficiency of the treatment plan.
3. Related Work
After analysing works related to this subject, weconcluded that the use of technology has shownsome proof of being more engaging and motivatingthan traditional materials. This leads to childrenpaying more attention to what they are learningand consequentially to better results.
Results obtained by measurements like A-B-A-B design, show that all sort of skills are improved,or at least maintained, when teaching is deliveredthrough this device. These findings are supportedby teachers and therapists observations.
This results are obtained not exclusively because
technology is being used but also because of factorslike personalization of the APPs, so that it adaptsto the child needs. Also, it allows us to concludethat reinforcement can be more effective when usingimages and sounds.
Although we would like to study more researchesthat were related to reading skills we find this re-sults are transverse, independently of the skills thatare taught. Given that iPads are a recent technol-ogy, and given the number of studies made aboutthis subject, there are not enough findings to con-clude if it can be more effective teaching one skillor another.
4. DevelopmentEnsina-me a Ler has the main objective of teach-ing children reading and writing skills, through aneffective and engaging interface. To develop thisAPP we used an agile approach well known in soft-ware development, the Incremental Development.This method is based on the idea of implementinga base version, expose it to users and evolve it basedon their comments.
4.1. Requirements analysis and definitionIn this phase, we defined what were the require-ments of this APP, what should it do and whatservices should it provide to reflect the needs ofits users and serve its purpose.To define them, weworked in collaboration with the therapist and nar-rowed it to the following:
4.1.1 Functional RequirementsFunctional requirements reflect how the systemshould react, behave and what should it providegiven a certain condition.  We set the followinglist as the main functional requirements list:
The user should be able to add words and im-ages.
The user should be able to turn reinforcementson and off.
The user should be able to choose between dif-ferent types of exercises.
The APP should have different types of exer-cises. It should cover reading, writing and as-sociation.
The APP should allow customization of bothexercises and interface.
The APP should allow to change between nor-mal an syllabic reading.
4.1.2 Non-Functional RequirementsNon-functional requirements are requirements thatare not directly connected to the services deliveredto the user but on which such services depend to
better perform their role. Non-functional require-ments relate to system properties, such as reliabilityand response time, and affect the overall architec-ture of a system.  Given that, the following re-quirements should be met:
The APP should have a clean interface.
The APP should have an intuitive interface.
The APP should allow to store data persis-tently.
The APP should easily allow to add and re-move data.
The APP should be scalable, when new data isadded efficiency should not be compromised.
The APP should allow to store severalwords/images without compromising effi-ciency.
The navigation between scenes should be easyand fast.
After establishing the previous requirements, we de-signed a prototype that allowed us to test our con-cept and provide specifications for the next step, forexample, of what should consist each exercise.
As mentioned before, the main purpose of this workwas to develop an APP that could help the acquisi-tion of academic skills, specifically reading and writ-ing skills. As we could conclude from related works,customization is the key to success. Not only it al-lows to engage the user visually but also to adaptthe difficulty of exercises to his level of knowledge.It is important for this APP that the addition andmanipulation of data 1 is easy and efficient. Thearchitecture has to be scalable, so that it is alwayspossible to add and remove new data without com-promising efficiency or efficacy.
5. Implementation5.1. Data Store
We decided to use Core Data framework to dealwith our data management.
The implementation of Core Data began whenwe had already started developing the first exercise,Leitura. In the beginning it was only possible toadd a word and its corresponding image. Otherattributes were added as needed.
To add new content, we built a settings menu,where we were able to add a new word and to seethe list of words already stored.
1In this APP we consider that data is words and images.
5.2. Leitura Exercise
The initial idea for the first exercise was to displayonly the word. The child would read it and whenthe therapist thought it was correctly spoken, thena reinforcement would be displayed, much like thetraditional way where cards are used. But we foundthat this exercise could be improved, and hopefullymore successful with a few changes. With this pur-pose a new attribute was added to our Palavras en-tity. We decided to include sound so it was possibleto give verbal prompts to the child. Now, when weadded a new word, we had to record the reading.
After implementing sound, we applied it to theexercise. This was made by transforming the worddisplayed on the screen into a button, so that eachtime this button was pressed it was possible to hearthe recording of the word. At the same time, weadded the corresponding image to the screen.
Both of this changes failed to succeed, because ifa child is exposed to the sound it repeats it withouthaving to look at the word. The same goes withthe image, the child identifies what it is supposedto say just by looking at the object displayed andnames it. While discussing this with the therapist,it was suggested that we displayed the image as areinforcement, which at the same time would helpthe child to make a correspondence between visualand written representation.
With a defined objective, we continued improvingour exercise. As suggested, we took the image, butwe still thought it would be of help to use the sound,so we went back to related works and related appli-cations to see if there was a way we could makesound useful in this context. As we searched, wecame across several applications used to teach emo-tions, which were never studied by us given that itwas a different subject and had a different purpose.This is well documented  as one common charac-teristic of children with ASD. Taking that into ac-count we decided to implement two buttons. A sadsmiley button (Fig.1(1)) and a happy smiley but-ton (Fig.1(2)). When the therapist considers thatthe child is not reading correctly it presses the firstbutton (the sad smiley) and the recording of theword can be heard, stimulating the child to repeatit and associating the feeling of sadness with fail-ure. When the therapist considers that the childcorrectly reads the word, it should press the secondbutton (the happy smiley) which will display theimage corresponding to the word.
The next step was to implement the syllabic read-ing. To do so, we implemented a new attribute toour entity Palavras and a new text field in wordadding. Also, a new settings screen, where it ispossible to turn on or off this feature, was added.
As we reached a solution that, from our perspec-tive, was fun and duly substantiated we decided
Figure 1: Reading Exercise.
to implement the reinforcement. To do so, we im-plemented a particle system, a rendering techniqueused in game physics and computer graphics thatallows to emulate a chaotic system, as fire, water,stars and others.
When the happy smiley button is pressed the im-age is displayed and this effect crosses the image,from left to right, with a falling star-like effect. Wealso wanted to make the effects customizable so wedecided to create a settings screen that allowed tochoose between four different particles. This diver-sity will also allow to keep the child interested inthe exercises.
5.3. Encaixe Exercise
Figure 2: Writing Exercise.
Next, we started implementing the second ex-ercise, Encaixe (Fig.2). This exercise is meant toteach children how to write a word. The first ap-proach to this exercise was to implement what wedecided in prototyping. We display a word dividedby letters, which we called tiles (Fig.2(3)), and wedisplay squares, called targets(Fig.2(2)), where tiles
are supposed to be dragged to form the word. Be-cause the tiles are shuffled, if a word is big, it mightnot be straight forward what it is supposed to bewritten, so we decided to use the sound featurehere too. When the word is displayed, the wordrecording is heard and at any time the speaker but-ton(Fig.2(1)) can be pressed to play the audio. Wealso implemented a visual prompt that makes itmore obvious that the tile was not placed correctly.When the tile is dragged to the wrong target itmoves away from its center. The reinforcement wasimplemented as in the previous exercise. When theexercise is completed, the reinforcement is displayedacross the screen.
We showed the exercise to the therapist and therewere several annotations. First, the task of drag-ging the tiles to the corresponding targets should bedone in order, it should not accept random placing.Second, when placing the tile it should be more ob-vious if it was done correctly or not. Third, whenthe tile is not placed correctly, there should be aprompt that indicated what was the correct tile.And fourth, the way the letters are displayed shouldbe customizable, it should allow to choose betweendifferent fonts and different capitalization styles.
To solve this problems we made a few changesto the exercise, except for the first problem, thatwas solved right away without major changes. Inorder to solve the second problem, of making thematching of tiles and targets more intelligible, weimproved the visual aid. We implemented a neweffect using our particle system. Every time the tileis placed in the correct target a explosion-like ef-fect is displayed. Though this visual reinforcementhelped, it was only applied when the correct matchwas made, so we decided to add sound indicationof what happened, in other words, we implementedsound to give positive and negative reinforcements.To do this, we added three different kinds of sounds:when a single tile is placed correctly, when a sin-gle tile is placed incorrectly and when all tiles areplaced correctly, solving the problem. The thirdkind of sound was also applied to the first exerciseand it is played at the same time as the visual re-inforcement.
The next problem to be solved was the displayingof a prompt when the tile was not placed correctly.It was implemented using a simple effect of scalethat will shrink and enlarge the target and tile(s)that are to be matched next.
Lastly, to solve the fourth problem, we imple-mented a new settings screen where we can choosebetween four different fonts and 3 different styles ofcapitalization.
5.4. Ligacao ExerciseThe third exercise (Fig.3) is intended to help chil-dren associate the word with its visual representa-tion, and its implementation is similar to the writ-ing exercise. In this exercise, the words are the tilesand the images are the targets. The child must dragthe words to the corresponding image which, likethe second exercise, will generate visual and soundreinforcements. Based on similar applications wehave a counter that increases the number of possi-ble matches on screen. At first, the exercise wouldstart with one match and at every three correct an-swers it would add one more, with a maximum offive possible matches. But, after presenting the ex-ercise to the therapist, we decided that it shouldstart with two matches. With only one match thechild would not have to read or understand whatwas presented to her. Knowing that the objectivewas to drag the tile to the target, she would sim-ply realize the automatized gesture without givingit any meaning.
Figure 3: Matching Exercise.
Since every child has a different learning curve,the therapist also suggested that it would be bet-ter if there was the possibility to choose with howmany matches it should start and how many cor-rect answers the child has to give to add a new one.With that in mind, we implemented a new settingsscreen. This screen allows to choose from two tofive matches to start, being that five is always themaximum of matches possible on screen, so that thewords do not become too small to read. It also al-lows to choose the number of correct answers neededto add a new match, with a range from one to ten.
5.5. ImprovementsWhen we reached this state of the APP, wherethe three exercises were implemented, we addedthe possibility of turning the reinforcements on andoff enabling the reinforcement schedule (see section2.1.1). Also, as we did with visual reinforcements,
we added the possibility of recording three person-alized sound reinforcements. This allows to use dif-ferent sounds and, for example, use a voice that isknown to the child to praise him.
The next step was to show it to the therapist, tohave feedback of the APP as a whole. She testedit for some time and reported to us details thatcould be improved. From the user perspective therewere some technicalities we had to solve. Whenadding a word, and respective image, we resizedthe image before saving it, but, if a image was big,it would still cut off some details. This was solvedby readjusting the size to which the image is scaled.
Another problem was that it was not possible toadd the same word with different images, a featurethat allows to understand the generalization of aterm. In other words, it allows to understand theword as a name to different representations of thesame object. It would always assume the first addedimage for different entries of the same word. Thishappened because we were saving the image withthe word as its name. To solve this we changed thesaving name of the image to the date and hour ofwhen it is added.
Regarding the words that were added, it waspointed out that as the list of words was gettingbigger it would become slower and less fluid andeven crashed one time. This happened because theimage displayed in the list was loaded at its full sizeand shrunk, overloading memory. The solution wefound to this problem was to save the image withtwo different sizes, so, by the time the user saves anew word, a thumbnail is also saved. This thumb-nail is used in the word listing and prevents thememory overload.
From the exercise perspective, Encaixe exercisewas being executed by trial and error, so we addeda model prompt. A model prompt is used to letthe child know what it has to do and, in this case,it consists of displaying in the targets the letterssupposed to be dragged (Fig.4). This prompt canbe turned on and off on settings.
Another problem, regarding every exercise, wasthat the words displayed were chosen randomly.Since there is a specific word that is worked at atime during therapy, while others are used as di-versions, the probability of the wanted word to befeatured on exercises was very small. To solve thisproblem we first had to implement the possibility ofchoosing a given word to be featured more, whichis done in word listing simply by touching the rowof the word.
Next, we implemented an algorithm that wouldallow us to solve this problem. In the Leitura andEncaixe exercises, every time a new exercise is dis-played, we create an array of three words. This newarray is filled with the selected word and two other
Figure 4: Model Prompt.
random words. We then choose a random positionof this array to be displayed in the exercise. Thiswill augment the probability of the selected wordbeing displayed to 33,(3)% .
In the third exercise, we used a different method.The chosen word is always displayed and, accord-ing to the number of matching pairs, we randomlychoose the rest of the words.
6. Evaluation and Results6.1. Evaluation MetricsSingle-subject designs are research designs, mostoften used in ABA research studies, in which thesubject serves as his own control, rather than usingindividual or group control.
Single-subject designs are used to monitor a sub-ject progress and test the effectiveness of an inter-vention.
This research has different designs composed bythe conjugation of different phases. Each phase canbe one of the following:
Baseline phase (abbreviated by the letter A)Refers to the status of the subject prior to in-tervention. Repeated measurements of the de-pendent variable are taken to serve as a controland enable to see the changes that occur withintervention. This phase is conducted withoutthe use of prompts or reinforcements.
Intervention or training phase (abbrevi-ated by the letter B) Represents the implemen-tation of the intervention. During this phase,repeated measurements of the dependent vari-able, using the same measures as before, areobtained to help determine if any change hasoccurred in comparison to baseline phase. Thisallows to determine if the intervention is effec-tive or not.
These phases are chained, one or more times, and
each conjugation, or design, allows to obtain differ-ent results. The possible combinations are:
The evaluation of our APP was made with the helpof the therapist and a male child. Typically thesetests are done with several subjects (e.g. three toeight) but they can be made with only one.  Inthis case it was only tested in one child due to lackof available children who would fit the criteria2.
6.2.1 Procedures and Experimental Design
The APP was tested with a male child of seven yearsold diagnosed with ASD (hereafter called tutee) andwith the therapist (hereafter called tutor).
The sessions took place in the tutee natural learn-ing environment, his classroom, at a table, whereboth trainer and participant would sit next to eachother.
Our APP was installed in an iPad2 and, given theavailability of both tutor and tutee, for the purposeof evaluation, it was only collected data from En-caixe exercise. We chose this exercise given that itrequires more interaction than Leitura exercise andmore reading/writing knowledge than Ligao.
The sessions were carried everyday, two per day,using the A-B-A design. Each session consistedof writing the same unknown word to the tutee,Foca.
For purposes of evaluation, every session con-sisted of five trials.
As explained before, baseline is meant to collectdata to serve as control and, in this case, determinethe ability of the tutee to write the word Foca.This phase only ends after four consecutive stablesessions.
This phase consisted of four sessions (two days)and, in each trial, the tutor would handle the iPadto the tutee and ask him to write the word. Basedon observation, the tutor would evaluate if the tuteehad given a correct or incorrect answer.
If the tutee tried to drag the wrong letter, or theright letter to the wrong target, the answer wouldimmediately be considered Wrong and a new trialwould be performed.
To better assess the capability of the tutee towrite the word, the exercise was presented withoutany prompt.
2A child diagnosed with ASD in the process of learningto read/write.
The intervention phase was applied to evaluate andregister the performance and evolution of the tutee.The duration of intervention was dependent of thetutee performance.
As in the previous phase, in each trial, the tu-tor would handle the iPad to the tutee and askhim to write the word. The first trial of the firstsession was made with the model prompt enabled(see Fig.4). If the tutee performed the exercise cor-rectly3 the answer would be considered a Right an-swer with prompt and in the next trial the tutorwould choose, based on the tutee performance, ifthe exercise should be displayed with or without aprompt.
If the tutee responded correctly, in the trial with-out the prompt, the answer was considered Rightand the next trial was performed in the same condi-tions. On the other hand, if the exercise was consid-ered Wrong the next trial would display once againthe prompt. This approach was followed through-out every session and every first trial of a sessionwould be based on the outcome of the last trial ofthe previous session.
Unless the first five trials were completed witha Right answer the tutee would perform a total oftwenty trials. Being that the the extra fifteen trialsare considered as training and not part of evalua-tion.
This phase ended when the tutee performed cor-rectly the first five trials of two consecutive sessions(i.e. one day sessions).
To determine if the intervention phase was success-ful we applied once again a baseline phase. Unlikethe other phases, that would start as soon as theprevious phase was finished, this new phase wasconducted only after three days. This is done tohelp determine if the effect of the intervention per-sists.
This phase had the duration of one day (two ses-sions) and was conducted like the baseline phase.The tutee had to complete five trials withoutprompts and, based on observation, the tutor wouldconsider the answers Right or Wrong.
6.2.5 Tutor Feedback
In order to evaluate the APP from the tutors per-spective we constructed a questionnaire (see Ap-pendix A) and conducted an interview with the tu-tor. This was done after the tutor completed theA-B-A research.
Figure 5: Results
6.3. Results6.3.1 Baseline
During baseline, the tutee never performed the ex-ercise correctly. Occasionally he would drag thefirst letter (F) to the correct target but no morethan that. Also, he never dragged more than oneletter correctly, or did it in two consecutive trials.
This leads us to conclude that the correct place-ment of the letter was random and that the tuteewas not familiar to this word or had any knowledgeof how to write it.
As we can see from Fig.5, the ability of the tutee towrite the word evolved throughout this phase. Al-though we present the complete results and datacollection, for the purpose of evaluation we onlyconsider the first five trials of this phase, in orderto have equal samples from different phases.
Reading the results (Fig.5), and considering al-ways the first five trials, we can see that in thefirst session of the first day (day 3), all trials hadthe prompt model enabled. Which tells us that al-though the tutee was able to give a correct answer,his performance was not good enough to presenthim the exercise without the prompt. By the sec-ond session, of the same day, his performance im-proved and he was given, once, the exercise withoutthe prompt but failed to give a Right answer.
On day 4, the second day of this phase, the firstsession was completed without changes on the tuteeperformance. It is on the second session of this daythat we could see a major improvement of his abilityto write the word. He performed the five trials withfive Right answers.
Although it was a good result, it can mean thathe memorized the answer from the previous session,and that is why we stipulated that there had to betwo correct sessions in one day two consider theword as acquired.
On the first session of the next day (day 5), and
3An answer is consider correct when every letter is placedin the correct order at the first try.
based on the outcome of the previous session, thefirst trial was presented without prompt, and thetutee failed to answer correctly. Only two of the fivetrials were completed with the Right answer which,although it was a step back from the previous ses-sion, also shows us that there was an improvementin comparison to the first session of the previousday.
The second session of this day was completed cor-rectly with five Right answers.
On the fourth day of this phase (day 6) the tu-tee performed both sessions correctly, the word wasconsidered acquired and the intervention phase wascomplete.
6.3.3 Follow-upReading the Results Graphic (Fig.5) we can see thatthe tutee was able to give four Right answers inthe first session and five Right answers in the sec-ond session. Given that, as mentioned before, thisphase was performed three days after the interven-tion phase was completed, we can conclude thatthe intervention was successful and that the tuteelearned how to write the word Foca.
6.3.4 Tutor FeedbackBased on the questionnaire we asked the tutor toanswer, we could draw several conclusions.
The interface of the APP was very adequate totarget users.
The amount and variety, as well as the type, ofthe exercises were also very adequate.
The tutee did not show any difficulty in per-forming the exercises and practically none inunderstanding what was the supposed type ofinteraction.
The tutee paid more attention and learnedmore easily to write a word using the APPwhen compared to the traditional methods.
The amount of prompts were sufficient butthere should be more reinforcements.
The visual reinforcements were much more rel-evant than sound reinforcements.
The level of possible customization of the APPis very adjustable to children with differentneeds.
The tutor considered that this APP was alsoeasy to use by herself.
She also considered that this APP could beused professionally as a learning tool and it wasbetter than most available APPs she had used.
In the interview, the tutor referred that this APPis very versatile and, in contrast to other APPs, hasthe advantage of having several types of exercisesbased on the same information. She also mentionedthat what makes this APP a good improvement,when compared to the existing APPs, is the factthat it is very customizable and allows the tutor tochoose what to display and when to display it.
As improvements she suggested that the enablingand disabling of prompts should be made on theexercise, or before an exercise begins, because, aswe can see in the tests made, it is necessary, fromtrial to trial, to turn them on or off and to do soshe always had to go back to settings to performthe action, which can make the tutee lose interestin the activity.
Also, referring to the visual reinforcements, andas concluded from the questionnaire, she mentionedthat at the begining, the tutee did not pay much at-tention to the reinforcement. She suggested, that itshould be possible to choose the amount of particlesthat are displayed, to create more or less impact ac-cordingly to the sensibility of the tutee to the visualstimuli.
The results of this study showed that, using ourAPP, the tutee made a rapid acquisition of how towrite the word Foca.
In 3 days the tutee went from absolutely noknowledge of how to write a word (zero correct an-swers) to complete a session with five correct an-swers. And, even after the intervention phase, with-out any help, he was still able do write it 90% ofthe times.
These findings suggest that the use of our APP iseffective as a teaching tool and are consistent withthe outcomes of the research we conducted previ-ously.
The main goal of this thesis was to fill a market gap.It had the objective of developing a tool that couldhelp Portuguese children diagnosed with a condi-tion of the Autism Spectrum Disorder. This toolshould help in the process of learning to readingand to writing skills and preferably make it easierand fun. Each children with such diagnostic hasspecial needs, that need to be addressed in differ-ent ways, so our tool had to be well supported bytheory and other works.
With that motivation, we created an iPad Appli-cation, Ensina-me a Ler, where most of the featuresavailable are based in concepts of Applied BehaviourAnalysis, a field of study well documented and withscientifically proven results. The research we con-ducted helped us understand how some of these con-cepts had been applied in teaching through technol-
ogy and if they were effective. After this review ofliterature, we defined what would be expectable ofsuch application by mimicking traditional materialsthat are used with the same objective. When wethough we had a solid foundation, we prototyped itand started implementing the application.
During implementation, and given that we choseto use an incremental design, we found that ourplanning had a complexity lower than what wouldbe required. All three exercises developed were, atthe end, different from what was initially planned.
To test if the achieved solution fulfilled our goals,we conducted a study with two possible users of thisapplication.
This research showed us that our application wasan improvement to traditional methods and thatthe concepts applied were also effective when deliv-ered through technology. The child showed interestin using the application and made a fast acquisitionof a new word.
The visual reinforcements proved to be captivat-ing, but should allow even more customization. Onthe other hand, the sound reinforcements did notwork as expected and the explanation we found forthis is that, although it is possible to customize thesound that is presented in a given situation, it isalways the same sound making it dull after severaldisplays.
Although the outcome of the tests were mainlypositive, this application should be tested withmore users. The fact that it was only tested in onechild does not allow us to conclude that it wouldbe an effective tool in every children, or in most ofthem.
7.1. Future WorkAlthough we consider this application successfuland that it fulfilled our goals, there are improve-ments and upgrades that can be made to this so-lution and that can make the process of learningreading and writing skills even more efficient.
On a first approach, the issues that were foundshould be corrected. For example, concerning thevisual reinforcements, maybe it would be a goodapproach to develop a mini-game where several ef-fects would be displayed and the child would in-teract with them. The effect that the child mostinteracted with would be the one displayed duringthe exercises.
To solve the lack of interest in the sound rein-forcements there could be the possibility to recordseveral audio files to the same action, that wouldbe displayed randomly.
Another issue that should be corrected is the lo-cation of the button that allows to enable and dis-able the use of prompts. It should be placed in ascreen, or position, of fast access avoiding the needto go through several screens to change it.
Also, regarding the actual solution, it should betested with more subjects in order to have enoughdata to conclude that this application was also ef-fective in other individuals.
One upgrade that we think would benefit thisapplication is the implementation of speech recog-nition in the Leitura exercise. This would makethe exercise less dependent of the tutor and, conse-quently, more captivating to the child.
Also, instead of choosing only one word, it couldbe possible to choose a variable set of words to workwith. This would allow to have a big collection ofwords and at the same time work one singularitypresent only in some words. For example, if thetherapist would like to work the pronunciation ofch, like in the words cho or chinelo, she couldchoose only the words that contained it.
Even though this upgrades suggestion would ben-efit this application, we consider that the first ap-proach should be to test the application with moreusers. This would allow to better assess the possi-ble issues of this solution before implementing newfeatures.
References Maria Luisa Rodriguez-Almendros Maria Jose
Martinez-Segura Alvaro Fernandez-Lopez,Maria Jose Rodriguez-Fortiz. Mobile learningtechnology based on ios devices to support stu-dents with special education needs. Computersand Education, 2012.
 American Psychiatric Association. Diagnos-tic and Statistical Manual of Mental Disorders.American Psychiatric Association, 2000.
 Jon Baio. Prevalence of autism spectrum dis-orders autism and developmental disabilitiesmonitoring network. Centers for Disease Con-trol and Prevention, 2008.
 Sathiyaprakash Ramdoss Mark F. O ReillyGiulio E. Lancioni Tonya N. Davis MandyRispoli Russell Lang Peter B. Marschikg DeanSutherland Vanessa A. Green Jeff SigafoosDebora M. Kagohara, Larah van der Meer. Us-ing ipods and ipads in teaching programs forindividuals with developmental disabilities: Asystematic review. 2012.
 Ofer Golan and Simon Baron-Cohen. System-izing empathy: Teaching adults with aspergersyndrome or high-functioning autism to recog-nize complex emotions using interactive mul-timedia. Development and psychopathology,2006.
 Autism Speaks Inc. How is autismtreated? http : / / www . autismspeaks .
org/family- services/tool - kits/100- day- kit/treatments-therapies. Accessed: 2014.03.28.
 PORDATA INE. Populacao residente se-gundo os censos: total e por grupo - por-tugal. http ://www.pordata .pt/Portugal/Populacao+residente+segundo+os+Censos+total+e+por+grupo+etario-2, 11. Accessed:2014.03.28.
 Leo Kanner. Autistic disturbances of affectivecontact. 1943.
 8th edition. (2009) Mosbys Medical Dic-tionary. Respondent conditioning.http ://medical - dictionary. thefreedictionary.com / respondent + conditioning. Accessed:2014.11.15.
 James Halle Gail Mcgee Samuel OdomMark Wolery Robert H. Horner, EdwardG. Carr. The Use of Single-Subject Researchto Identify Evidence Based Practice in SpecialEducation. 2005.
 Ian Sommerville. Software Engineering.Addison-Wesley, 9th edition, 2011.
IntroductionBackgroundAutismABA - Applied Behaviour AnalysisABA and Autism
Related WorkDevelopmentRequirements analysis and definitionFunctional RequirementsNon-Functional Requirements
ImplementationData StoreLeitura ExerciseEncaixe ExerciseLigao ExerciseImprovements
Evaluation and ResultsEvaluation MetricsSingle-Subject Designs
EvaluationProcedures and Experimental DesignBaselineInterventionFollow-upTutor Feedback