I've been continuing my phonological analysis of Old Tirina, and after a lot of faffing about in a spreadsheet, here are some conclusions I've come to:
Theoretically, Old Tirina has geminated consonants. Practically, there are so few words that include them that they might as well not exist. I am giving serious thought to doing away with them for future words.
I treat Old Tirina as if it doesn't have diphthongs, but realistically, it absolutely does. In particular, it has a bunch of Və vowel sequences that surely must be the result of some kind of vowel breaking or reduction in the past. I need to revamp the sound changes to Modern Tirina to handle these in a more interesting (and more consistent) way. There's not a ton of long vowels in Old Tirina either, and I wonder if I can connect these two.
I need to use the glottal stop more frequently and more interestingly.
Consonant clusters make no sense. There's /tr kr gr/ but no /pr br dr/? Yet there's also /mr/? I need to figure out what clusters really are permissible. While I'm on the subject, I need to figure out where Modern Tirina clusters came from (where only /pr mr fr/ are acceptable). Syllable-final clusters are also weird: it appears basically any rC is valid, but I also have a bunch of /nd/ plus one instance of /ʃt/. Are any other ʃC clusters valid? What about nC?
It appears bilabial consonants can't appear in a syllable coda, with the exception of /m/. Why is /m/ an exception? I'll probably leave this as an irregularity and not try to explain it, as it's quite common in OT.
By far the most common syllable is /a/. (70 occurrences in my OT lexicon! Next most common was /i/ at 36.) A solid quarter of syllables had no onset and 55% had no coda. Of syllables with onsets, the most common were /d s g/. Of syllables with codas, the most common were /n m r/. Weirdly enough, the most common syllable with both an onset and a coda is not /dan/, which occurred just once in my data, but /ben/.
Finally, you probably should have an excuse prepared when a coworker stops by your desk and asks what on earth you're working on, as "I'm doing phonological analysis of a fictional language" isn't likely to yield a good response. I went with "character frequency stuff", myself.