Romance etymology: A computerized model

Abstract

DESPITE THE FACT that the majority of sound changes within a given geographical area, within a given span of time, and within a given sound environment have long been considered regular and reduceable to definite patterns of change, and despite the fact that these changes have almost always been noted as variations of "If, then" statements, there has not yet been, to my knowledge, an attempt to program these patterns and statements into a computer-generated etymological dictionary of regularly derived cognate forms a dictionary which could surpass in scope and accuracy any etymological dictionary yet produced, and one which offers the promise of being an important instructional tool. Such a dictionary would differentiate between regular formations and those forms which, for whatever reasons, did not follow regular phonological changes, and could note what the expected derivation would have been. The intrinsic value of the study lies not in the computerization of phonological changes occurring in the development of a Romance language, per se though that is, in itself, a worthwhile accomplishment. The real value lies, rather, in the possibilities afforded by the expansion of the program. At present, however, the program is still in its infancy: morphological and analogical variations to normal development have not been programmed; discriminations in the exact stages of development are not precise, with output printed arbitrarily as Classical Latin, Vulgar Latin, and Old French; the development is assumed to be that of the Francian dialect, though some exceptions may be found. The program as it stands traces possible word-histories without considering the effects of morphology, analogy, retardation or other factors which alter "regular" or expected soundchange patterns. This, to me, seems a sufficient beginning. Many of my colleagues suggested that I tape and store the contents of the etymological dictionaries alrea y in existence, those which had been produced by years of laborious and tedious tracing by hand, but the program which I envisioned was to serve as a model for a program which could, conceivably, generate a reconstructed ProtoI doEuropean sound base as well as all the regularly derived cognate forms for all PIE languages and their dialects. Such a program would provide scholars with non-attested formations; with formations that have not survived or with variants which