I'm that way, too (writing words first, and then writing the tune), and generally, I try to model the melody on the sounds of the words as spoken -- that is, I speak the line out loud, as if I were having a spirited conversation with somebody (as though I were trying to convince them of the truth of my argument -- not as though I were reciting a poem for an English class), and listen for when my inflection goes up or down, and then write the melody to match that. As for the duration of each note, I try to match the spoken words, too. For example, in the word "into", it takes longer to say the second syllable than the first, so I might make the first syllable an eighth note, and the second syllable a quarter note.
The first verse and chorus give me the melody. Then I sing that over and over, and write the subsequent verses as though I were writing a filk.
It's a system that works well for my ear, and produces songs that I can sing easily while I putter 'round the house (which is when I do most of my singing). Which is why I find it completely baffling and frustrating when my Dad informs me that the words and notes don't match up. And this is true for every song I've shared with him.
It's like we're speaking two different musical languages, and I don't know how to translate.
And I haven't gotten any "second opinion" feedback on the matter to figure out whether or not it's my ear and & brain combo that is out of kilter. Though I do belong to a mailing list dedicated to singing and sharing rounds, and I've gotten feedback on the melody to a round I'd written [the overlap among the parts led to a chord progression that shifted keys]. But no one said that the words and notes didn't match up. Part of me assumes that's because the words and tune did match up. I can never be sure, though, if the others on the list had the same problem my dad does, and just didn't want to say anything. ...I really wouldn't be surprised if it was my brain, though, since visible symptoms of my CP (out of whack muscle tone, coordination and balance) are controlled in the same part of the brain that processes the experience of rhythm. And when my dad was a kid, he took music lessons as a drummer, so his ear is really attuned to rhythm. But if there is a test for "rhythm deafness", I've never taken it (being able to clap along at a song circle is not a skill that's rated with great importance by the mainstream medical community, unfortunately).
The above long ramble has been a set-up for the following question:
Is there some formula for matching notes to syllables that I can use as a back-up for when my own ear may not be reliable? Should I, for example, base duration of notes not on the length of a syllable as spoken, but on its accent (In which case, "into" would get a longer note on its first syllable, instead of its second)?