zompist bboard

It seems that, in a root which ended in one of /m n l r/ (I'm not sure about /w j/), this consonant became syllabic when the root was in zero-grade, even if followed by a vowel. Thus a root like ter could have had a derived form tr.-o-s, with syllabic /r/, when I'd expect simple monosyllabic tros. Which of the following is most likely to be true?

1. bricka is mistaken, and this never actually happened.
2. "bricka, it's really obvious why that happened, and you shouldn't have to ask".
3. "Well, you see, bricka, it's a bit more complicated than that...", followed by a helpful explanation.

And what happened with /w j/, if that's not a silly question?

There may have been a schwa before (or after) the vowel until a fairly late stage of PIE, which would put it in the same category as English with words like "fur" > "furry", which doesnt change to /fri/ because it's analyzed as two separate morphemes.

I've never heard of this, but it sounds similar to the Sievers-Edgerton Law. This law effects all sonants and basically says that a sonant is vocalized after a long vowel or pause. This occurs across word boundaries, as well as within a word. It only effects sonants in word initial clusters if the word is monosyllabic.

It seems to be established pretty well for Germanic and Vedic for /j/ word internally (Sievers original formulation), but the extension of the law by Edgerton has been heavily criticized in recent times.

Edit: Thinking about it a bit more I can think of two examples where this vocalization could take place. If the root is of the form CeRH (which might be hard to distinguish from CeR) then in zero grade we'd have CRH and the R would vocalize (though the details would depend on individual languages). Also, I'd expect the N in #CNV to vocalize, though I don't know if there's any established law for this.