Fulani-Arabic Transliteration
Click for sample image.
Click for sample image.
Introduction
A computer software project that I undertook has left me in possession of some technical development information and programs that I feel may well be of use to translators in some parts of the world. I would like to make this kernel of knowledge freely available to any interested linguistics workers. The software is a ‘quick fix’ solution to a rather curly problem as explained below. However, the ideas contained therein could prove valuable to some people and may warrant further development.
History
About two years ago someone asked me if I could help with a computer problem. That person was Dave Wilkinson, a missionary currently working on Bible translation in Burkina Faso and who was home on furlough in Nowra on the south coast of New South Wales, Australia.
The problem was presented to me as a question; is it possible to utilise a computer to automate the process of transliteration?
Understanding what transliteration was and enough about programming to know that a computer solution was possible, I believed that there would almost certainly be commercial software available for the task. When I asked Dave if he had made any inquiries along these lines he said that no one he had contacted was able to offer any help except to suggest engaging a professional programmer. Dave also said that the programmers he approached either felt they did not have the necessary language skills or were too busy to undertake any new projects.
I couldn't believe this and after further discussion Dave unveiled some additional background information.
The language Dave is translating is called Fulani (Fulfulde). It's written form has only recently been developed using a phonemic system incorporating Roman scripting, in compliance with internationally agreed standards. The problem is that most Fulani people have virtually no reading ability.
However, Dave had an idea that could potentially accelerate the process of acquiring reading skills. The majority of Fulani are Muslim and the men of this religion are required to learn passages of the Koran. They don't need to be able to understand the Arabic (in the novice stages), just recite it. Once a person learns the basic sounds of the language and some Arabic script rules they can sound out the words, similar to reading music. This is possible because Arabic is phonemic. So it’s not actually a language reading skill, but a sound reading skill they’ve acquired. Of course this ability is a prerequisite to actually learning to read, but the point here is that the process takes time, often quite a long time. Dave’s idea was that because a large number of Fulani men already possessed this basic skill and since both Fulani and Arabic are phonemic, why not transliterate the Fulani script into Arabic script. Then, by sounding out the Arabic transliteration, the "reader", and others listening, would be able to hear their own Fulani language.
There are a few phonemes in Fulani (Fulfulde) that don’t occur in Arabic so, to accommodate these, Dave had to create new Arabic symbols using minor modifications of existing ones. After a few trials Dave realised that the system worked well. He transliterated the book of Luke by hand which took about six weeks. It was a laborious and error prone task; hence the original question; is it possible to utilise a computer to automate the process of transliteration?
The Project
Well, a colleague and I managed to achieve just such a solution. It took about three or four months to complete and we didn’t need to learn Arabic.
Having only limited skills in professional programming languages like C, and relatively little discipline in others, I soon realised that in order to complete the project, before Dave and his family returned to Burkina Faso, I needed some assistance. My friend Graham Kettlewell agreed to help. He even lowered himself to my level by doing most of the coding of the transliteration engine in QBASIC (using my sketchy algorithms) so that I could read and debug the program.
The complete process includes the following operations:
- Open an existing Fulani source file. These files were created by Dave using a JAARS editor and include embedded formatting commands.
- Pre-process this file to eliminate some format inconsistencies.
- Create an output file containing the Arabic transliteration and implementing the scripting rules of Arabic.
- Implement the formatting commands and convert the transliterated file to Rich Text Format (RTF).
- The above four steps are automated by separate modules under the control of the master program and take less than a minute to complete, depending on computer speed.
- Import the RTF file into a Windows 3.x or Win95 word processor and apply any further enhancements or formatting corrections manually.
- Print the document.
- Prepare mirrored image masters for offset printing.
- The above three steps take between 3 and 20 minutes to complete depending on the length of the files.
It would be appropriate to explain at this point that, because of the ‘quick fix’ nature of the project, some shortcuts were taken during development. These include:
- Use of third party shareware text filter program (may require permission, but it’s function could easily be incorporated into a professionally developed project).
- Use of less than optimal performance compilers in two versions of BASIC
- Modification of an existing Arabic font for Windows (may require permission or the design of purpose built fonts from scratch).
- My implementation of formatting commands in RTF is not comprehensive and may require further work if necessary.
- Use of a standard (Western: left to right) word processor or publishing program necessitated the creation of a mirrored Arabic font. This shortcut greatly simplifies the editing of Arabic. The only real drawbacks are that you must get used to viewing the Arabic script this way and some form of reversal or mirroring must be applied (usually photographically) before creating the masters for production printing.
On the positive side, the program is fairly easy to maintain and can be easily modified for different source languages because of the dual lookup table design. In general, I feel that the program demonstrates a potentially valuable technique rather than advocate any particular style.
The purpose of freely releasing the programs and related data is specifically for further development and the assistance of language workers.
Please contact us for additional information.
Ken PercivalResources for Download
Useful Links
SIL - Computers and Writing Systems Microsoft - Transliteration Utility v1.0 (TU)
UPDATE NEWS February 2009
- Development terminated.
- Project status RETIRED.
says that you are visitor number