The Compansion Project

The Compansion technique is a syntactically and semantically-based AAC method that generates complete well-formed sentences from a user input of uninflected content words. It has applications both as a rate enhancement technique for people who use word or picture based systems (it reduces the number of keystrokes and the number of word forms needed) and as a tool for people learning language who do not yet have good grammatical skills.

Table of Contents Purpose Method Progress Results Publications Acknowledgements

Contributors Pat Demasco, Yu Gong, John Hughes, Mark Jones, Kathy McCoy, Chris Pennington, Stephanie Snyder, Wendy Zickus

The goal of this project is to increase the communication rate of people with severe disabilities via natural language processing techniques. We have developed a technique called Compansion which expands a compressed (telegraphic) sequence of words (input by the user) into a semantically and syntactically well-formed utterance. At the same time, we wish to do this by placing as little a burden on the user as possible. Thus, we are not interested in a simple coding system where sentences have been stored and are indexed by their content words.

Input from the user will be the roots of the content words of the desired utterance; thus, many function words including determiners (e.g., the, a) and prepositions (e.g., of, in) will normally be left out. The system is responsible for filling in missing words as well as correctly conjugating the verb and forming a syntactically correct utterance. We attempt to form an utterance whose word order most closely reflects the word order given in the original input string. For example, if the system is given "APPLE EAT JOHN," we would like the system to produce the sentence, "THE APPLE IS EATEN BY JOHN."

The Compansion project has been divided into four phases. The syntactic parser phase accepts input from the user and enhances it to designate modifier/clause attachments, compound noun structures, and in some cases, word sense disambiguation (sometimes resulting in multiple output structures). In the next phase, the semantic parser, the output from the syntactic parser is placed into a well-formed semantic structure. This phase is responsible for determining what role (semantic relationship) is played by each word of the input. To illustrate, in the example above, the parser recognizes that EAT is a verb that prefers an animate ACTOR and an inanimate/food OBJECT.

The semantic structure built from this phase is passed to the translator phase which is responsible for determining how each piece of the semantic structure can be realized in English. The translator is responsible for adding language specific information to the semantic structure and for translating that structure into a representation appropriate for the sentence generation phase. The sentence generator uses a functional unification grammar to generate an English sentence from the specification constructed by the translator.

A full prototype of the Compansion system has been completed in Common Lisp and tested. Although the range of syntactic forms is not exhaustive, Compansion does handle declarative sentences that include multiple adjectives, prepositional phrases, possessive noun phrases, direct and indirect objects, and some verbal clauses (e.g., "I want to go to the store"). Questions, imperatives, complex verb tenses, and do-support are also provided. The present system has a vocabulary of over 1000 words and has the capability to infer the verb or subject in some situations. In addition, Compansion has been augmented to understand some metaphorical verb uses. A graphical interface emulating a word-based communication device has also been designed to demonstrate Compansion at conferences and for visitors at ASEL.

Discussions with the Prentke Romich Company and Semantic Compaction Systems have centered on taking some of the technology developed in this project and transferring it into the development of an intelligent therapy tool. Such an application provides us with a limited domain that allows us to fully specify the lexical knowledge required. Part of the technology developed in this project has been implemented in C++ which makes it available for quick and efficient transfer to future applications.

McCoy, K. F., & Demasco, P. (1995) Some applications of natural language processing to the field of augmentative and alternative communication. In Proceedings of the IJCAI '95 Workshop on Developing AI Applications for Disabled People (pp. 97-112). Montreal, Canada.
[abstract], [text (48K)], [postscript (158K)]

Pennington, C. A. (1995) Providing intelligent language feedback for augmentative communication users. In Proceedings of the IJCAI '95 Workshop on Developing AI Applications for Disabled People (pp. 125-136). Montreal, Canada.
[abstract], [text (30K)], [postscript (95K)]

Jones, M. A. (1994) Transparently-motivated metaphor generation (a Ph.D. dissertation). Technical Report 94-29, Department of Computer and Information Sciences, University of Delaware, Newark, DE.
[abstract], [postscript (690K)]

McCoy, K. F., Demasco, P. W., Jones, M. A., Pennington, C.A., Vanderheyden, P. B., & Zickus, W. M. (1994). A communication tool for people with disabilities: Lexical semantics for filling in the pieces. In Proceedings of the ASSETS '94. Marina del Rey, CA.
[abstract]

McCoy, K. F., McKnitt, W. M., Peischl, D. M., Pennington, C. A., Vanderheyden, P. B., & Demasco, P. W. (1994) AAC-user therapist interactions: Preliminary linguistic observations and implications for compansion. In M. Binion (Ed.), Proceedings of the RESNA '94 Annual Conference (pp. 129-131). Arlington, VA: RESNA Press.
[abstract], [text (13K)], [postscript (75K)]

Jones, M. A. (1993).Transparently-motivated metaphor generation (a Ph.D. proposal). Technical Report 93-14, Department of Computer and Information Sciences, University of Delaware, Newark, DE.
[abstract], [postscript (351K)]

Demasco, P. W., & McCoy, K. F. (1992). Generating text from compressed input: An intelligent interface for people with severe motor impairments. Communications of the ACM, 35(5), 68-78.
[abstract], [text (50K)], [postscript (125K)]

Jones, M. A. (1992). Generating a specific class of metaphors. In Proceedings of the 30th Annual Meeting of the Association of Computational Linguistics (pp. 321-323). University of Delaware, Newark, DE.
[abstract]

Jones, M., & McCoy, K. F. (1992). Transparently-motivated metaphor generation. In R. Dale et al. (eds.), Automated Natural Language Generation, 6th International Workshop on Natural Language Generation (pp. 231-246). Berlin / Heidelberg: Springer-Verlag.
[abstract]

Jones, M., McCoy, K., & Demasco, P. (1992). The problem of metaphor in intelligent AAC systems. In J.J. Presperin (Ed.), Proceedings of the RESNA International '92 Conference (pp. 363-365). Washington, D.C: RESNA Press.
[abstract], [text (14K)], [postscript (79K)]

Demasco, P., McCoy, K. F., Jones, M., Pennington, C., & Snyder, S. (1991). Compansion: A technique that applies natural language processing to augmentative communication. Rehabilitation R & D Progress Reports / Journal of Rehabilitation Research and Development, 29, 180-181.
[abstract], [text 4K], [postscript (51K)]

Jones, M., Demasco, P., McCoy, K., & Pennington. C. (1991).Knowledge representation considerations for a domain independent semantic parser. In J.J. Presperin (Ed.), Proceedings of the Fourteenth Annual RESNA Conference (pp. 109-111). Washington, D.C: RESNA Press.
[abstract], [text 14K], [postscript (79K)]

Demasco, P., McCoy, K.F., Jones, M., Pennington, C., & Stum, G. (1990). Rate enhancement through sentence compansion. Rehabilitation R & D Progress Reports / Journal of Rehabilitation Research and Development, 28, 197.
[abstract], [text (3K)], [postscript (49K)]

McCoy, K., Demasco, P., Jones, M., Pennington, C., & Rowe, C. (1990). Applying natural language processing techniques to augmentative communication system. In H. Karlgren (Ed.), Proceedings of the Thirteenth International Conference on Computational Linguistics (pp. 413-415). Morristown, NJ: Association of Computational Linguistics.
[abstract]

McCoy, K., Demasco, P., Jones, M., Pennington, C., & Rowe, C. (1990). A domain independent semantic parser for compansion. In J.J. Presperin (Ed.), Proceedings of the Thirteenth Annual RESNA Conference (pp. 187-188). Washington, D.C: RESNA Press.
[abstract], [text (14K)], [postscript (61K)]

Demasco, P., McCoy, K., Gong, Y., & Pennington, C. (1989). Towards more intelligent AAC interfaces: The use of natural language processing. In J.J. Presperin (Ed.), Proceedings of the Twelfth Annual RESNA Conference (pp. 27-30). Washington, D.C: RESNA Press.
[abstract], [text (11K)], [postscript (61K)]

McCoy, K., Demasco, P., Gong, Y., Pennington, C., & Rowe, C. (1989). A semantic parser for understanding ill-formed input. In J.J. Presperin (Ed.), Proceedings of the Twelfth Annual RESNA Conference (pp. 35-37). Washington, D.C: RESNA Press.
[abstract], [text (8K)], [postscript (57K)]

McCoy, K., Demasco, P., Gong, Y., Pennington, C., & Rowe, C. (1989). Toward a communication device which generates sentences. In J.J. Presperin (Ed.), Proceedings of the Twelfth Annual RESNA Conference (pp. 41-43). Washington, D.C: RESNA Press.

This work has been supported by a Rehabilitation Engineering Center grant from the National Institute on Disability and Rehabilitation Research. Additional support has been provided by the Nemours Foundation.