Lexical Sorting in Snobol

Alphabetical sorting routines useful in processing natural language texts are explored in this article. Four basic algorithms (Bin, Bubble, Tree and Chain sorts) and five SNOBOL implementations of these algorithms are discussed. In addition, the authors deal with problems of program-defined special character sets and sorting orders diverging from standard machine collating sequence, as well as questions of producing reverse-alphabetized word lists. Comparative running times for these programs on representative corpora of varying sizes are tabulated and compared. For those readers unfamiliar with SNOBOL, the article offers a brief introduction to the language sufficient to allow understanding of these and many other SNOBOL programs. Thus the programmer who works with other languages can both familiarize himself with SNOBOL, and at the same time learn enough to translate the sort programs into his usual language. Reference is also made to SPITBOL, a recently developed high-speed compiler for SNOBOL.