Reconstruction talk:Proto-Indo-European/-ih₂

Morphology

@Rua The Erlangen paradigms do not work very well in this kind of suffixes. This suffix is mostly added to stems, not to roots. That means we cannot use a three-morpheme paradigm and because of this it fails. I see you are trying to make sense out of the morphology but the model simply cannot explain it. For example, this suffix has a genitive in *-i̯áh₂as, instead of *-i̯áh₂s as the model predicts, we know it must have been *-i̯áh₂as for the same reason we reconstruct *-ohₓom in the genitive, because Vedic has an hiatus that must come from a "VHV" PIE sequence. Vedic preserved this suffix in two forms, a tonic and an atonic one. The suffix never shows mobility, but instead is always static, sometimes in the root, others in the suffix itself. All this predicted by the compositional model. But paradigmatic models fail terribly trying to explain a word with three full-grades (gen.sg ǵénh₁tri̯ah₂as). Most of them give a lot of excuses, they mostly rely on generalizations but the zero grades are never attested. They are evidently too biased to visualize the holes in their theories. It would be great if we could add the compositional model along with the paradigmatic one. --Tom 144 (𒄩𒇻𒅗𒀸) 22:02, 2 February 2018 (UTC)

What's the compositional model? —Rua (mew) 22:15, 2 February 2018 (UTC)

@Rua: In short terms. It rejects paradigms, and instead classifies morphemes into three groups, inherently accented morphemes, inherently pre-accented morphemes, and neuter morphemes. Now, using three rules it derives the shape of a word. To exemplify, let's see the gen.sg of *ǵénh₁trih₂. It's morphemes are ǵénh₁-ter-i̯éh₂-és, the "*e" in "-ter-" is deleted because of the zero grade rule, that gives us ǵénh₁-tr-i̯éh₂-és, and now all accents are deleted except the right-most one (BAP), giving /ǵénh₁tri̯ah₂as/. This matches perfectly the Vedic evidence. It can also be used to explain aberrant nouns such as *méh₂tēr. It is way more accurate since it does not need to make up unattested full-gades just to keep the theory in one piece (e.g. almost all i and u-stems, see *suHnús), and does not need to delete others without evidence (e.g. gen.sg "*ǵn̥h₁tr̥yéh₂s"). I could give you a more detailed explanation if you want, or you could go directly to the source. --Tom 144 (𒄩𒇻𒅗𒀸) 22:40, 2 February 2018 (UTC)