Patterns of lexical loss from Latin to Romance
This dissertation is essentially a contrastive study of two groups of Latin words, those which are identifiable as etyma in the Romance languages, herein termed survivals; and those which did not evolve into Romance words, herein called failures Three distinct approaches to the data are utilized in this study: (1) a morphological analysis to determine which form classes in Latin evince traits conducive to survival or failure; (2) a phonological analysis to determine what characteristics favor survival or failure; and (3) a semantic study to determine the extent to which meaning affects a word's probability of failure or survival The corpus consists of all Latin words in Cassell's New Latin Dictionary. Survivals are identified in W. Meyer-Lubke's Romanisches etymologisches Worterbuch. In the corpus there are 10,791 failures and 4,057 survivals Each word was encoded for machine analysis according to a series of criteria: survival density, i.e., the number of major and minor Romance languages in which the word survives; traditional morphological classification (declension or conjugation); whether simple or compound, whether case or tense limited; number of syllables; position of stress; vowel and consonant sequences; semantic field; number of non-synonymous definitions; whether pejorative, ameliorative or neutral; and finally its relative frequency of occurrence in a body of Classical Latin literature was noted Chapter I, which treats of morphology, proves fairly conclusively that form classes strong in Latin remain strong in Romance. Second conjugation verbs, adjectives, adverbs and conjunctions have relatively high losses, but many were low-frequency words in Latin Chapter II, on phonology, has yielded disappointingly inconclusive results. Only word length emerges as an important feature. Words of up to three syllables have high rates of retention in Romance, but words of four or more syllables have high rates of failure Chapter III, semantics, is rich in statistical tendencies. Words belonging to semantic fields dealing with nature, vegetation, animals and the dwelling place have favorable rates of retention, whereas more abstract semantic fields such as quantity, mind and thought, social and political relations, religion and warfare have poor retention rates