John Muccigrosso (@JD_PhD), whose published work was so helpful in the creation of the DCC core Latin vocabulary, alerted me recently to a hidden gem of classical scholarship, George B. Hussey’s Latin Homonyms: Comprising the Homonyms of Caesar, Nepos, Sallust, Vergil, Terence, Tacitus, and Livy (Boston: Benj. H. Sanborn, 1905; available here in scanned form). It is a very full list of homonyms from these authors (the entire corpus in most cases, but only [only!] the orations of Cicero and early books of Livy, up through XXII), crucially including citations of each instance. So, for example, we learn that feri means “fierce” 6 times in that body of Latin, and “strike!” only once. The string ferias means “holidays” twice, and “you may strike” once.
Fascinatingly, Hussey chose to include at the bottom of the page “unmated” homonyms, that is, cases in which one of the pair, while grammatically possible, does not actually appear in the corpus he examined.
It is as if Hussey in 1905 had chosen to do precisely the task that computers are notoriously poor at doing. The potential here for natural language processing in Latin, and the improvement of automatic parsers, seems manifest. If we could expand on Hussey’s data to take in a larger corpus, and continue to use human inspection to categorize the occurrences, then we could get a pretty good picture of the likelihood of a particular homonym deriving from a particular dictionary headword, even without looking at the context. With his homonym list, finding thousands more occurrences of each is now a trivial task. With the help of contextual analysis of the kind being pursued by Patrick Burns (@diyclassics) and others, we could really make progress. For me the goal here is to augment tools such as the Bridge, which creates vocabulary lists that are accurate and helpful to readers. Combining this sort of data with good, author-specific dictionaries holds enormous potential to ease the burdens of Latin and Greek learners in the years ahead.