As part of my larger project (QLTP), one of the things I’ve been working on in the last few months is software to divide a Latin word into syllables, determine which ones are short, which long by nature, and which long by position, and find the word-accent. Here is my test module, analyzing the 123 words of Horace, Carmina 2.7. It has one bug and couple of refinements still to be added:
- The bug can be seen in the second-to-last word: it’s marking a last (or only) syllable short when it is actually long by position. That’s just defective logic in the code and needs further analysis.
- I haven’t yet added code to look for prefixes: if it is a compound of ob (etymologists seem to be unsure), then 91 o·blī·vi·ō·sō (˘ ̄ ˘ ̄́ ̄ ) should be ob·lī·vi·ō·sō ( ̄ ̄ ˘ ̄́ ̄ ), and the same goes for 74 obligatam and (mutatis mutandis) 62 sustulit.
- Latin dictionaries don’t seem to bother with syllabification, but it’s not quite so unproblematic as that implies. For instance, Gildersleeve and Lodge (§ 10) say that MN “under Greek influence . . . belongs to the following vowel”. I don’t doubt that (e.g.) Polymnestor would be divided Po·ly·mne— rather than Po·lym·ne—, but is it really true that somnus would be so·mnus rather than som·nus, and amnis a·mnis rather than am·nis? If so, I’ve been pronouncing them wrong for decades. The same problem comes up with ST. Can anyone point me to more recent work on this?
- Latin editors who distinguish consonant V from vowel U seem to put some of them consistently in the wrong category. Whether QU should be QV doesn’t really matter, since the combination is a special case metrically and needs to be handled as such, but what about the Us in suavis, anguis, and sanguis, and the second U in unguentum? Aren’t those all consonants? They’re not listed as diphthongs in any grammar I’ve seen, and my software is currently misdividing unguenta (word 97) as four syllables (un·gu·en·ta) when it’s actually three (un·guen·ta). So why aren’t they spelled angvis, sangvis, svavis, and ungventum? That’s how they’re pronounced, and syllabized. More urgently, where can I find a complete list of these exceptions? The class does not include every Latin word in which NGU is followed by a vowel, because relanguit (for instance) is four syllables, not three.
Comments and questions will be very much appreciated. I plan to offer my syllabizer as a stand-alone module, not just a part of the larger project. As soon as I get this module working correctly, I will add code to search for elisions and scan whole lines.