Thanks! I'll have a look.

I feel like there's enough people doing this that there must be a set of shared assumptions, but I don't think I've ever seen them spelled out. 4/4

Like, what does autosegmental GEN or rich base look like? How do correspondence and underspecification work? (In all fairness, it wasn't that clear how underspecification worked in 'normal' autosegmental theory). Can markedness and faithfulness both refer to autosegments and association lines? 3/n

I sometimes run into abstracts or papers that just pop autosegmental representations into tableaux, and I feel like I'm supposed to know how this works. But I really, really, really don't. 2/n

At some point when I realized that lots of seriously hardcore developers and CS faculty just stick with plain html, I decided to do the same. It's universally readable on any device any time and will survive a nuclear war. Good enough.

Thanks! I'll take a look.

Ooh, I gotta take a look at that.

Yeah, regardless of those fiddly details, I think the overall suggestion is on the mark: you could probably extract a large amount of stress info from those two properties. Shortening of immediately post-tonic consonants is also an underrated stress cue for American English.

For vowels, if the transcriber is using schwa instead of, e.g. wedge or lax I, they probably already know the stress and these transcriptions might just reflect that knowledge instead of anything phonetic. For some tokens. But you could probably get a lot out of those transcriptions.

If it's just presence or absence of aspiration, you'd have a hard time distinguishing stress from word-initiality. Word-initial voiceless stops preceding schwa are lightly aspirated, pre-stress ones are heavily aspirated, there's probably overlap in distributions of VOT values.

Except Finnish. You go to your room, Finnish, and think about what you've done.

Gulp

As a longtime employee of a third-tier public flagship in flyover country, which is broadly representative of where the majority of American college students study, it was *maddening* to read the endless think pieces about how Yale, Oberlin, and Harvard reveal what's wrong with American education.

I've sort of started going in this direction as well, trying to limit my example sets to things from really well-documented primary sources or my own data. But it feels like it will take 100 years to replace *all* of my teaching materials this way. I guess I should keep plugging.

www.youtube.com/watch?v=oBmb...

I had not seen that! Thank you.

I feel similarly. Although I think in my case I ended up in a weird position in between phonetician and phonologist, which didn't do me any favors for the first 14 years of my career.

There's definitely some overlap with the Port and Leary critique, basically that the IPA is not fit for purpose. I agree with that for a lot of things. But then beyond that level there's even more noise entering the channel, because capsule summaries often ignore exceptions and complexity.