oa A Cheremebased Sign Language Description System

Because sign languages are not written, it is challenging to describe signs without knowing a spoken language equivalent. Sign language does not represent in any direct way the form of the spoken language either by visually representing sounds or syntactic sequences of words of the spoken language. One sign may mean an entire Arabic phrase and vice versa. Sign language can only be described, animated or videotaped. For example, a Deaf person may have it difficult to convey a sign for a concept such as “moon” on paper if he/she does not know the Arabic word. In this paper, we describe a notation system enables users (among other functions) to choose. Sign Language users identify the four cheremes for each hand for the STEM sign for which they want to find Standard Arabic equivalents by using pictorial lists. Using the four descriptors (hand shape, movement, location, and palm orientations of both hands, Deaf and Hard-of-Hearing users can describe Moroccan Sign Language (MSL) signs and find corresponding Arabic STEM terms, MSL and Arabic definitions, and concept pictures. The program will then search the database for the sign that most closely matches the selected cheremes in the STEM Sign Database. Then Standard Arabic information (definitions, parts of speech, etc.) and MSL information (graphic signs and videos) are displayed to the user. There are two possible scenarios that occur: if the system finds an exact match of the selected cheremes, then it will return the exact sign with Standard Arabic and MSL information; otherwise, it will display signs that most closely match the selected cheremes. In our database, signs and words have an N-N relationship, which means that there are signs that refer to multiple Standard Arabic words and vice-versa. Therefore, we had to group the database by signs and try to select signs with more than one Standard Arabic equivalent. Our database is in alphabetical order by Arabic base form word/stem. This means that when there is an Arabic word that has more than one meaning and consequently different signs, there are separate entries for that word. In order to reverse the strategy, that is, identify signs that can be expressed as different Arabic words and invariably have different meanings, we have reordered the database according to sign graphic file name. By programming retrieval using this reverse strategy, we have created the first-ever digital MSL thesaurus.The creation of this resource required, among other things, the identification and development of codes for the MSL cheremes, code assignment to the STEM signs and addition of Arabic definitions and videotapes of MSL translations of definitions. A usability and feasibility evaluation of the tool was conducted by having educators of deaf children, their parents, and deaf children themselves test the software.