An ICT-based borderless combination of alternative and augmentative communication systems.

Abstract

In working with people with communication and/or language special needs and to facilitate their communication and social integration, alternative and augmentative communication (AAC) systems are used. Many special needs persons have improved their communication skills combining signs from different systems. Using this approach we propose to combine several AAC systems using new information and communication technologies (ICT). Our system helps instructors teach, and students learn several AAC systems. It is designed for gestural sign languages and graphic sign systems. It contains a tool kit to assist instructors in the generation of bi- and tridimensional communication boards. Tales and games present great psychological and educational benefits. Therefore, our system, HAPPY, includes several tools to work with any tale or game based on AAC systems. Both of these enjoyable activities help instructors introduce new vocabulary and strengthen the assimilation of the AAC systems that are in training. HAPPY helps instructor evaluate basic user capabilities, such as memory, attention, perception, psycho-motor skills, and is useful in producing personalized user working environments "ease-of-use" and "ease-of-learning". HAPPY helps instructor ensure technical accessibility and great usability, ease of learning, and high performance for special needs users.

**********

But will he/she speak?

When parents are told about their children's disabilities or limitations, they usually ask in an anguished voice: "But will he/she speak?" This concern for their children's future is really related to their communicational ability, that it is undeniably essential to their social integration in the human community which is mainly based on communication and particularly based on spoken languages. Inhabitants of the disability communities must be helped by building awareness around the most visible areas that could block their social development and integration. We must avoid negative thoughts that turn differences into impediments. We must emphasize abilities, not limitations. The subject matter in this article centers on people with communicational disabilities, people who live in apparent silent communities. Surely, if parents have got something to say, they should ask a better question: "But will we be able to communicate with him/her?"

We bear three key goals in mind: help "able-bodied" people learn to hear these sounds of silence, help special needs people develop communicational skills and, particularly, spoken-based abilities if they are present, and help any kind of people learn to communicate to each other. The system here proposed, HAPPY, fits in with these goals. On the one hand it helps "able-bodied" people learn or strengthen specific communication systems, on the other hand HAPPY is designed to increase special needs people communicational skills by helping instructors evaluate basic user capabilities, and by using multi-lingual and multimodal reinforcement educational models constructed by instructors. Finally, HAPPY makes the first move towards the third goal when it helps instructors train people in some of the different communication systems that are unknown to them.

In working with people with communication and/or language special needs and to facilitate their communication and social integration, alternative and augmentative communication (AAC) systems are used (Lloyd, Fuller and Arvidson, 1997; Glennen and DeCoste, 1997). Alternative communication (AC) systems are intervention instruments often used to teach, through specific procedures of instruction, a structured set of non-vocal codes. In alternative and augmentative communication (AAC) systems the term augmentative refers to the fact that the teaching of AC systems has among its goals to lead to spoken language. They support those who have difficulties understanding and being understood. In addition, they ensure an alternative communication system if the person does not ever speak.

A better knowledge of the AAC systems used by special needs persons facilitates their communication and social integration. If parents, relatives, friends and instructors of a special needs child have a better knowledge of the AAC systems the child uses, then they will be able to promote AAC-based conversations more frequently. Therefore, they are actively participating in the rehabilitation of the child. Really, any AAC system should be a part of a global programme, which is necessary to resolve communication and language problems. One of the programme's key goals must be to improve the social-cognitive and social-communicative aptitudes.

Undoubtedly, the best way to succeed in improving and strengthening what we have learned about a communication system is the participation in communication acts with other people who use the same system. The interlocutors in these communications acts may be physically present or not. We center our thoughts on the Internet and on Virtual Reality (VR). We cannot forget the social impact of the Internet and Information Technology (IT). Special needs persons have many problems with the Web's client-server interface. The Web Accessibility Initiative (WAI), launched in April 1997, is pursuing accessibility of the Web through several areas of work: technology, guidelines, tools, education, and outreach, and research and development.

The ideal solution seems to be an adaptive Web's client-server interface, according to the different disability communities. That is, a true personalization where Web design ensures accessibility for every user by adapting to the user's preferences. Nevertheless, Web personalization is still in very early stages of development (Paciello, 2000). An alternative way arises: we could open specialized communication channels within the Internet allowing one to use different AAC systems in Web-based communication acts. On the other hand, there are positive experiences of immersion into virtual reality, which are highly beneficial to learning. Virtual Reality enables instructors to design, create and structure the simulated physical and social environment into which the AAC system user lives, so they can improve his/her social relations. He/she can even learn certain strategies in accordance with his/her competence level, social development and level of control. In one of these experiences (Schroeder, Cleal and Giles, 1993), teenagers learned to build virtual worlds: future worlds, contemporary worlds and historical worlds, and navigate through them. At Shepered School, in Nottingham (UK), they constructed a collection of 3D virtual reality simulations based on Makaton and British Sign Language (BSL). The interaction of the students with the 3D VR simulated scenarios improved their capacities of incorporating and retaining new knowledge. According to these positive precedents, we could incorporate a tool kit to our system. This tool kit helps instructors develop some scenarios and games, where the user is a virtual actor who can select any communication system with the other actors.

These virtual environments, which will be accessed via the Internet, will add a new dimension to chats. They also stimulate the "augmented Internet" including distance communication using different AAC systems.

Who/What is HAPPY?

Surely many of us have been frustrated at not being able to communicate with another person. We decided to call HAPPY to our system mainly due to the unhappiness feeling product of this frustration. But, where there's a will there's a way. HAPPY is an information and communication technology (ICT) system whose implementation is nearing completion. Although due to its nature, it is an "always-in-progress" system. In short, it helps instructors teach, and students learn several AAC systems. Some goals involved in HAPPY project are:

* The teaching/learning of the Spanish sign language (SSL), although the same could be done with another sign language.

* Attempt to extend HAPPY to other gestural and graphic sign systems, such as PIC, PCS, Bliss, Rebus, etc.

* To be the support of a language where any data could be represented, regardless of its vocal or non-vocal origin.

* To be a module of wider projects. For example, in a special needs person evaluation and rehabilitation project, where HAPPY provides instructors with information about the user's communication and motor skills.

* To help identify the barriers that special needs people must overcome in order to use the Web. Thus, to achieve accessibility and usability to the greatest extent possible.

* To stimulate the "augmented Internet" including distance communication using different AAC systems.

* Among some relevant features helping to achieve the previous goals, we point out the following. HAPPY,

* helps instructor construct multi-lingual and multimodal models that are useful for learning. These models should include signs where meanings are context or user dependent. Using these models, instrutors could plan either a "lesson" or a whole "course" or strategies for learning signs of greater conceptual difficulty for the user.

* allows instructors to combine words from several semantic fields and from any considered communication system. Vocal systems are a priority, but in the absence of them, HAPPY automatically uses an orthographic system.

* works under a multi-modal user interface framework. HAPPY should include conventional input devices such as joysticks and coded or unencoded keyboards, and less traditional input devices such as headmice, gloves or tracking systems based upon video or electro-myographic information.

* gives feedback to the user. With full-feedback option selected, whenever the user makes a manual sign, chooses a graphic sign, or says a word, HAPPY automatically reproduces it using all the AAC systems implemented.

* is open with respect to the multilingualism, the plurality of meanings and to the multimodality, i.e. it is possible to add or modify communication systems, contents, semantic categorizations and channels of interaction.

* recognizes non-standard signs and some signs made by persons with certain motor problems. If an AAC system is in training, then according to the progress of the user, HAPPY will hint at the possibility of modifying the percentages of occurrence of signs that do not belong to the AAC system that is in training.

* is a reinforcement learning tool assisting instructors that provides user and instructor with intrinsic feedback consisting of information about user's performance. This feedback complements the necessary extrinsic feedback provided by instructors, which consists of any extra information added for training purposes, such as emotions, words of praise or encouragement.

* includes the most used graphic sign systems: PIC, PCS, Bliss and Rebus. Instructors could include others, such as pictograms SYMS and OAKLAN, PREMACK cards, lexigrams and several SIGSYM systems.

* contains Board-Builder, a tool kit to assist instructors in the generation of bi- and tridimensional communication boards. Board-Builder helps instructors:

* configure the spatial distribution of signs in any board as they see fit.

* simplify the signs which are more difficult for users by changing their shape. Another option is to use some adapted vocabulary, or signs from other communication systems. Instructors could use supplementary drawings to facilitate the learning and understanding of the signs, such as the "help lines" (Warrick, 1978).

* predict next signs in the speech.

* includes several tools to work with tales and games based on AAC systems (e.g. "sign twisters"). Both of these enjoyable activities help instructors introduce new vocabulary and strengthen the assimilation of the AAC systems that are in training.

* could incorporate a Virtual Reality tool kit which helps instructors develop some scenarios and games, where the user is a virtual actor who can select any communication system with the other actors. These virtual environments, which will be accessed via the Internet, will pursue the "augmented Internet" with different AAC systems.

* can perform management tasks such as joins, withdraws, reports or simple statistics.

Assisting instructors through reinforcement learning

Although not used by psychologists, the term reinforcement learning (RL) has been widely adopted by the artificial intelligence (AI) community. The simplest RL methods are based on a common-sense idea: if an action is followed by a satisfactory state of affairs or an improvement in the state of affairs, then the tendency to produce that action is strengthened (Barto, 1998). The actualization of the state of affairs is based on imitation. If an AAC system is in training, in accordance with a prescribed learning model designed by the instructor, HAPPY helps the user learn or strengthen this specific AAC system if necessary. But in order for people to be able to imitate something, it is necessary for them to have acquired imitation as a meta-ability. An illustrative example is given by Iacono and Parsons (1986), in which the gestural sign teaching by imitation is compared with a hand-guided strategy. The results suggest that a hand-guided method is better. Because of this, it is necessary to define some strategies to achieve this meta-ability. For example, with respect to gestural signs, a special needs person needs to be able to identify the parts of the hand, at different conceptual levels, fingers, palm, back, wrist, phalanges, etc. One possibility is to try to match the user's hand with the synthetic hand, which can both seen on the screen. This helps to avoid a possible and unlucky initial mirror effect. At this stage tasks are simple, for example to extend a certain finger. The set of tasks is automatically generated by HAPPY, which also evaluates if the mirror effect has been removed after the user's hand has been gradually taken away. HAPPY examines imitation meta-ability proficiency from simple to complex tasks following the ideas of Berges and Lezine (1965). Once an established proficiency threshold has been exceeded, teaching/learning assistance stages can begin.

Notice that, as far as we know, HAPPY is the first system to combine the following features: on-line assistance to adjust a sign to a theoretically synthesized perfect sign, recognition of the user's sign, evaluation of a distance between both signs, identification of the widest separated sign constituents couples and collaboration closely with instructors on the definition of an strategy of on-line adjustment based on visual matching. This adjustment is similar to what is usually done in vocal rehabilitation computer applications such as SpeechViewer of I.B.M. Moreover, there are specific modules that measure the accuracy of a sign articulation and other modules that propose examples of discrimination tasks between signs of similar articulation.

Thus, the key role of HAPPY is to be a reinforcement learning tool assisting instructors. Some dimensions in this assistance are explored below.

Experience with deaf people using bilingual educational models has been positive (Tetzchner, 1984). In such a model, sign language is considered as the native language and the oral one is viewed as a second and foreign language. If there is a loss of motor skills, non-manual signs are necessary. Many special needs persons have improved their communication skills combining signs from different systems. Therefore, HAPPY, is multi-lingual. The information is arranged into different lexical categories and monothematic sections. In any communicational act instructors could combine words from several semantic fields and from any considered communication system. Thus, we can say that HAPPY possesses a plurality of meanings attribute. Nevertheless, we always have to bear in mind the goals of homogeneity and stability of the linguistic input. Vocal systems are a priority, but in the absence of them, HAPPY automatically uses an orthographic system.

One long-term goal within HAPPY project is the definition of a multi-modal user interface (MMUI) where the Window Icon Menu Pointer is another mode and not a paradigm. It should include traditional input devices such as mouse, trackball, joystick, data tablet, coded or unencoded keyboard, wall/desktop/portable touch panels, microphone, image scanner and CCD television camera; less conventional devices such as foot and headmice, gloves, suits, electromagnetic, acoustic, mechanical or electrostatic 3D trackers, tracking systems based upon video, electro-optical, electro-myographic, electro-encephalographic or electro-oculographic information, eye movement tracking systems from infrared, etc. Actually, HAPPY accepts, besides these conventional devices, the CyberGlove and the 3D tracker IsoTrack II. Automatically, it accepts multi-modal simultaneous inputs. This helps to clear up possible uni-modal ambiguities.

Another way of resolving such ambiguities is to give feedback to the user. Whenever the user makes a manual sign, chooses a graphic sign, or says a word, HAPPY automatically reproduces it. The system uses a digitized voice and digitized/synthesized manual signs to supply hearing and gestural accurate feedback to the user, while verifying the input at the same time. Optionally, this feedback can be disabled or enlarged with graphic sign productions. HAPPY possesses an open system architecture, with respect to the multilingualism, the plurality of meanings and to the multimodality. That is to say, it is possible to add or modify communication systems, contents, semantic categorizations and channels of interaction.

Lastly, it is user dependent. In addition, HAPPY can perform management tasks such as joins, withdraws, reports or simple statistics. It is capable of learning morpho-syntactic and motor dimensions from the user. For example, after necessary previous training, it recognizes non-standard signs and some signs made by persons with certain motor problems. This is based on previous studies (Leon-Rojas, Moreno, Silva and Morales, 2000; Leon-Rojas, Masero and Morales, 200I). Therefore, HAPPY is able to adapt to user's circumstances. Please imagine an AAC system to be trained and a user whose vocabulary consists of signs of several AAC systems. According to the progress of the user, HAPPY will hint at the possibility of modifying the percentages of occurrence of signs that do not belong to the AAC system that is in training.

HAPPY helps instructor construct various kinds of models that are useful for learning. These models should be multi-lingual and multimodal, and they should include signs where meanings are context or user dependent. Instructors construct a learning model for each user. The use of a suitable learning model could improve user's performance. HAPPY provides user and instructor with intrinsic feedback (Schmidt, 1982) consisting of information about user's performance. If HAPPY is being used without the presence of instructors, all information about user's performance is saved and will be reported to instructors when they come back. Nevertheless, we think that the most valuable feedback is extrinsic feedback (Schmidt, 1982) provided by instructors. This extrinsic feedback consists of any extra information added for training purposes, such as emotions, words of praise or encouragement. So the key role of HAPPY is to be a reinforcement learning tool assisting instructors.

Creating and using communication boards

HAPPY includes the most used graphic sign systems by Europeans: PIC (Pictogram Idiogram Communication), PCS (Picture Communication Symbols), Bliss and Rebus, and, as immediate extensions, pictograms SYMS and OAKLAN, PREMACK cards, lexigrams, and several SIGSYM systems. Instructors could include other graphic sign systems: iconic pictograms SYMS and OAKLAN, the original non-iconic PREMACK cards, their iconic extension (Deich and Hodges, 1977), the modification by Carrier (1974) with colour coding and the lexigrams (Romski, Sevcick and Pate, 1988). As far as we know, there is no standard for the SSL with respect to the SIGSYM signs (Tezchner and Martinsen, 1992). Nevertheless, there are SYGSYM signs for the American Sign Language (ASL) (Cregan and Lloyd, 1984) and for the British Sign Language (BSL) (Cregan, 1982).

HAPPY contains the GS-2DBoardBuilder tool to assist instructors in the generation of bidimensional communication boards (classic-style), and the GS-3DBoardBuilder to assist instructors in the generation of tridimensional communication boards. The latter are based on some ideas of 3D environments for file management (Santurino, 1995). These boards constitute direct intervention proposals, since they try to improve the user's spatial conception. It is possible to have vocal inputs and outputs. The 2D boards can be printed. In the 2D case, the distribution of signs automatically obeys the classic adaptation of the method of Fitzgerald (1954). In a 3D board, our proposal is that the more a sign is used, the nearer and higher it is placed with respect to the user. This idea is similar to keyboards with interchangeable keys: it is well-known that the Dvorak keyboard has a lead over the QWERTY design. The reason is that the former places vowels and other high-frequency characters under the home positions of the fingers, i.e. where the latter places ASDFGHJKL. However, instructors could configure the spatial distribution of signs in any board as they see fit. Communication boards with hidden information have produced encouraging experience. The appearance of a (2+1/2)D board is similar to a 2D board, but behind each sign there is another hidden sign. Instructors choose these couples of signs according to user discourse. The spatial complexity of a 3D board with hidden signs helps instructors understand user's spatial-memory skill. At the same time, by performing these tasks, always under a suitable learning model defined by the instructor, the user improves his/her spatial-memory skill by using HAPPY.

Some people cannot differentiate among certain graphic signs, for example, if they suffer from stimulus overselectivity (Loovaas, Koegel and Schreibman, 1979). In this case, instructors can simplify the signs which are more difficult for them by changing their shape (Tetzchner and Martinsen, 1992). The GS-xDBoardBuilder tools deal with this possibility. Another option is to use some adapted vocabulary, or signs from other communication systems. The kernel of the GS-xDBoardBuilders contains SignForecast, which is very useful for persons with limited vocabulary and motor problems. The 2D builder contains an orthographic board generator and WordForecast, which predicts the next letters and words. In addition, it is possible to use "help lines" (Warrick, 1978) with these builders. These lines are certain supplementary drawings that instructors could use to facilitate the learning and understanding of the signs. These boards are displayed on a wall, desktop or portable touch screen. The user can access each sign by pointing to it, by using a manual scan or a guided scan, which often uses sequential rules and/or categorical groups. The way the scan is made depends on the user abilities and on the previously agreed arrangement of the signs. It is also possible to activate the earlier intrinsic feedback options.

Self-expression through games and tales

Once an instructor has defined a learning methodology, HAPPY can help him/her to correct some mistakes and to reinforce learned subjects. In order to do this, it uses conversational games, based on phrases that are generated from template sets defined by the instructor. Some of these templates can be offered by HAPPY if a user is not able to complete the task. Although we have thought about how these games relate to gestural signs, they can also be extended to graphic signs. As examples, we mention the following:

* Splitting up: given a sign, the user has to identify its articulatory components (handshape, orientation, position, kind of movement and movement direction).

* Further variants:

(v. 1) find and make some signs under certain constraints (for example, signs starting at the head, or ending with a certain handshape, or that include a circular movement, or with constant handshape, etc.);

(v.2) given an image, find and make the greatest number of signs concerning the objects in the image, under some restrictions, similar to the previous point;

(v.3) for each articulatory, component of a given sign, find mud make another sign that includes it;

(v.4) the same as the last one but trying to make a sentence with all the signs. Other variants present more constraints: the sign must belong to a certain grammatical category (verb, noun, adjective, etc.) or to a concrete semantic category such as animals, communication means, etc.

* SynAntSigns: given a sign, find and make synonyms and antonyms.

* Chainings: given a sign, find and make another sign so that the last handshape of the first sign and the first handshape of the second sign are equal.

* Linguistic puzzles: make sentences from a given set of signs. There are two main tasks here: to make the signs accurately one by one, and to achieve a sentence which is constructed correctly both syntactically and semantically.

* Sign twisters: equivalent to tongue twisters.

* Rhymes: construct several sentences so that the last signs of each sentence are as similar as possible.

In addition to games, tales also present great psychological and educational benefits. Both of these enjoyable activities, tales and games, help instructors introduce new vocabulary and reinforce the assimilation of the AAC systems that are in training. In the examples above the AAC system in training is based on gestural signs. The fact that a special needs user has the opportunity, and desire, to tell a tale, means we can strengthen his/her social integration. HAPPY can generate, sometimes automatically, mono-thematic boards for each tale. In this case, the signs will be ordered following the narration (Borras, 1998). However, instructors could order the signs as they see fit. The user may also choose to make a new story with signs used in a certain given tale.

Identifying user skills

An important part of user training consists of developing a working knowledge of basic user capabilities, such as memory, attention, perception, psycho-motor skills. This knowledge is useful in producing personalized user working environments that do not place excessive demands on the user's abilities which we mentioned earlier. Gestural signs are essential to evaluate the user's motor skills. As a general procedure, HAPPY investigates the 12 handshapes and 6 movements proposed by Dennis, Reichle, Williams and Vogelsberg (1982). The analysis made by HAPPY on the articulations of these configurations provides the instructor several clues about which signs will be the most difficult to articulate and how to simplify them. HAPPY helps instructors identify the signs that may produce greater difficulty in articulation for the user. HAPPY also helps instructors recognize the articulatory components that cause these difficulties. Using ideas based on previous studies (Leon-Rojas et al. 2000, 2001), deviations to and from a stereotype * generated from knowledge and experience of multiple experts * are computed. This also facilitates the automatic generation of certain kinds of games like, for example, sign twisters. In general, it is essential for an AAC system to evaluate the user's skills. HAPPY can help the instructor in several ways because it:

* performs a useful role as an interlocutor with syntax-free or simple-syntax-based communication systems. In this case, a semantic field and a vocabulary must be set up. Generally, this is done using a controlled language (Hill, 1972).

* assists user in the evaluation of some imitation tasks such as exact sentence repetition (syntactic imitation), or repeating the meaning of a sentence (semantic imitation).

* helps instructors discover minimal contrast sign couples, which allow him/her to develop future user discrimination tasks between those signs in the same couple.

* collaborates with instructors in locating some relevant moments have occurred, such as hesitations, self-adjustments, unintelligible productions, interruptions, elided or incomplete signs, unfinished productions, pauses, etc. Therefore, HAPPY assists in the transcription of the sample for its posterior analysis by programs such as SALT (Miller and Chapman, 1987) or CLAN (MacWhinney, 1991).

Conclusions

HAPPY is designed for a large group of special needs people. It works with them under a wide range of conditions. Universal accessibility and universal usability paradigms are always present in our thoughts. Notice that technical accessibility is necessary but not sufficient for usability of a design. Even if a design is theoretically accessible because it follows the technical accessibility standards to the letter, it can still be very hard to use for people with disabilities (Nielsen, 2000). Our interfaces were designed according to some metaphors related to real objects in users' environments. This is the case of a communication board, which is a real object in the environment of some users. In addition, Virtual Reality enables us to design, create and structure the simulated physical and social environment into which the AAC systems user lives. An immediate consequence of this interaction is the improvement of user's social abilities. We must completely trust in the special needs person's potentials. We must avoid negative thoughts that turn differences into impediments. We simply must target aid at people who need it. We merely must help special need persons develop their potentials. Universal accessibility and usability, as well as social integration of vulnerable groups are nothing like easy goals but they are compulsory ones. HAPPY aims to be one small step in these directions. Certainly, one is driven to keep on researching.

Acknowledgements

We would like to acknowledge and thank Dr. S. A. Welch for mentoring us and especially for her thought-provoking remarks. We also would like to acknowledge and thank Amy Shrock for revising our article. She modified both grammar and syntax and contributed to making the Spanish article more understandable in English.

References

Barto, A. G. (1998). Reinforcement learning. In M. A. Arbib (Ed.), The Handbook of Brain Theory and Neural Networks (pp. 804-809). Cambridge, MA: The MIT Press.

Berges, J. and Lezine, I. (1965). The Imitation of Gestures: A Technique for Studying the Body Schema and Praxis of Children Three to Six Years of Age. London: William Heinemann Medical Books.

Borras, P. (1998). Roser: Proceso de integracion en la escuela ordinaria para una alumna que utiliza sistemas aumentativos de comunicacion. In C. Basil, E. Soro-Camats and C. Rosell (Eds.), Sistemas de Signos y Ayudas Tecnicas para la Comunicacion Aumentativa y la Escritura (pp. 173-186). Barcelona: Masson.

Carrier, J. K. (1974). Nonspeech noun usage training with severely and profoundly retarded children. Journal of Speech and Hearing Research, 17, 510-517.

Cregan, A. (1982). Sigsymbol Dictionary. Cambridge: LDA.

Cregan, A. and Lloyd, L. L. (1984). Sigsymbol Dictionary: American Edition. West Lafayette, IN: Purdue University.

Deich, R F. and Hodges, P. M. (1977). Language without speech. London: Souvenir Press.

Dennis, R., Reichle, J., Williams, W. and Vogelsberg, E. T. (1982). Motoric factors influencing the selection of vocabulary for sing production programs. Journal of the

Association for Persons with Severe Handicaps, 7, 20-32.

Fitzgerald, E. (1954). Straight Language of the Deaf. Washington D. C.: Volta Bureau.

Glennen, S. L. and DeCoste, D. C. (1997). Handbook of Augmentative and Alternative Communication. San Diego, CA: Singular Publishing Group.

Hill, I. D. (1972). Wouldn't be nice if we could write computer programs in ordinary English--or would it? The Computer Bulletin, 16 (6), 306-312.

Iacono, T. A. and Parsons, C. L. (1986). A comparison of techniques for teaching signs to intellectually disabled individuals using an alternating treatment design. Australian Journal of Human Communication Disorders, 14, 23-34.

Leon-Rojas, J. M., Moreno, J., Silva, A. and Morales, M. (2000). A family of divergences between Phi-probabilistic sets with application to handshape recognition. In F. J. Ferri, J. M. Inesta, A. Amin and P. Pudil (Eds.), Advances in Pattern Recognition (LNCS 1876) (pp. 777-786). Berlin-Heidelberg: Springer-Verlag.

Leon-Rojas, J. M., Masero, V. and Morales, M. (2001). Bayesian inference of cheremes (handshapes) using type 2 fuzzy sets. Paper presented at the XXVI Congreso Nacional de Estadistica e Investigacion Operativa. Ubeda: University of Jaen (In Spanish).

Lloyd, L., Fuller, D. and Arvidson, H. (1998). Augmentative and Alternative Communication: A Handbook of Principles and Practices. Boston, MA: Allyn & Bacon.

Loovaas, O. I., Koegel, R. I. and Schreibman, L. (1979). Stimulus overselectivity in autism: A review of research. Psychological Bulletin, 86, 1236-1254.

MacWhinney, B. (1991). The Childes Project. Tools for Analyzing Talk. Mahwah, NJ: Lawrence Erlbaum.

Miller, J. and Chapman, R. (1987). SALT: Systematic Analysis of Language Transcripts, User's Manual. Madison: University of Wisconsin Press.

Nielsen, J. (2000). Designing Web Usability: The Practice of Simplicity. Indianapolis: New Riders Publishing.

Paciello, M. G. (2000). Web Accessibility for People with Disabilities. Lawrence, Kansas: CMP Books.

Romski, M. A., Sevcick, R. A. and Pate, J. L. (1988). Establishment of symbolic communication in persons with severe retardation. Journal of Speech and hearing

Disorders, 253, 94-107.

Santurino, M. C. (1995). Administrador de Programas: HABITAT. Unpublished degree thesis in Computer Technical Engineering, University of Extremadura, Escuela Politecnica, Caceres, Spain,

Schmidt, R. A. (1982). Motor Control and Learning. Champaign, IL: Human Kinetics.

Schroeder, R., Cleal, B. and Giles, W. (1993). Virtual Reality in Education: some preliminary social science perspectives. Paper presented at the Interface to Real and Virtual Worlds Conference, Montpellier.

Tetzchner, S. Von (1984). First signs adquired by a Norwegian deaf child of deaf parents. Sign Languages Studies, 44, 225-257.

Tetzchner, S. yon and Martinsen, H. (1992). An Introduction to Sign Teaching and the Use of Communication Aids. London: Whurr.

Warrick, A. (1978). Blissymbols for Preschool Children. Toronto: Blissymbolics Communication Institute.

Juan-Miguel Leon-Rojas, University of Extremadura, Spain

Valentin Masero Vargas, University of Extremadura, Spain

Maria Montana Morales Morgado, Universidad Laborai, Spain

Leon-Rojas is Associate Professor in Applied Mathematics at the UEX from 1991. He has conducted 17 postgraduate courses in Computer Science. Masero Vargas is Assistant Professor in Computer Science and Engineering. Also, he graduate student at the Minimally Invasive Surgery Center (Spain) and at the Computer Science and Medical Engineering Departments of Wake Forest University, USA. Morales Morgado is Assistant Professor at the Higher Comprehensive School "Universidad Laboral" and coauthor of one book about office automation.