A.2.1 General T
As there were, to the best of my knowledge, no programs available in Australia that could perform the specific analyses that I required, they had to be created for the purpose 95 . There were two ways that this could have occurred.
The first way would have involved having a trained programmer write them under my supervision. I would have specified the task, and the programmer would have devised a program that would have analyzed the text in appropriate ways, and thus obtained the information. This would have provided potentially limitless possibilities for breakdowns in communication, distortion of the task, misguided efforts, and the like.
The second way was to write the programs myself. This second course also had disadvantages, although these were outweighed in this case by the advantages. Had I needed to learn to program a computer, the balance probably would have swung the other way. The chief disadvantage of writing my own programs was the time it took. The advantages for me were the ability to specify precisely what I wanted done, the flexibility this entailed, the chance to quickly follow up leads, and the lack of the communication gap between researcher and programmer.
The programs were written in Pascal for several important reasons. Firstly, it has a number of features, such as flexible data-structures, adequate string-handling, reasonable input/output, and a wide variety of control statements, that make it appropriate for this sort of application. Secondly, the version of the language available on the University computer, a Digital Electronics Corporation VAX 11/780, supporting VAX/VMS, was error-free and fast. This meant that programs took minutes, rather than hours to run. Fourthly, I knew PASCAL already, unlike other possible languages such as LISP and SNOBOL, which being interpreted are necessarily slower, and which I would have had to learn from scratch. Fifthly, VAX/VMS Pascal supports a wide range of non-standard file structures which I used to speed file access.
All the programs were desk-checked. A piece of test-data were devised to simulate all the foreseeable conditions that might occur in the text (included at the end of this appendix as section A.3) The programs were then run, and the results obtained compared with those calculated by hand for the same piece of data. Only when I was satisfied that the programs were functioning flawlessly to the limits of the test data, did I use them on the whole text.
[ Back to Site Index / Roadmap ] [ Table of Contents ] [ E-mail Author ]