From what could I derive a morpheme that explicitly marks a noun as being a phrase head?

Question

I have a number of languages I want to combine into a macrofamily, and one thing they have in common is nouns that ending the the pattern *-(V)nVm, e.g. *-n-om in a PIE-esque language, -unum in an Akkadian-esque language, etc. This really looks more like two morphemes smooshed together, and since *-om is a core argument case marker in the PIE-esque language, *-(V)n- is presumably something else that isn't a case marker, but that can stack with a case marker. Like what?

One idea I thought was cool was to use this *-n- to indicate that the noun is a phrase head. Semitic languages have something similar with the "construct state", which marks the possessee, i.e. the head of a possessive phrase, e.g.

*gi     iki
 1.SG   father
"I am a father"(?)

gi     iki      -n
1.SG   father   EA
"my father"

But my idea generalizes this to other noun phrase types, like a noun modified by an adjective or locative expression:

*es      geneš   -a
 fish    canal   LOC

es     -in   geneš   -a
fish   EA    canal   LOC
"the fish in the canal"

*ilu     mulkud
 white   pearl

ilu     mulkud   -un
white   pearl    EA
"a white pearl"

Or a noun modified by a relative clause:

giš     g     -ab    ua    -du
house   1.SG   ERG   see   PST
"I saw the house"

giš     -en   g      -ab   ua    -du
house   EA    1.SG   ERG   see   PST
"the house that I saw"

ukku        -a    eguš        -ar   deb   -ek
saltwater   LOC   blue_crab   PL    sit   3.PL.S
"there are blue crabs in the water"

ukku        -n   -a    eguš        -ar   deb   -ek
saltwater   EA   LOC   blue_crab   PL    sit   3.PL.S
"the water in which there are blue crabs"

I'm glossing the *-(V)n here as EA for état d’annexion "annexed state", a term used in grammars of the Shilha Berber language for a vaguely similar marking.

This begs the question of where the hell this EA morpheme evolved from, since not all of the macrofamily has it. The World Lexicon of Grammaticalization doesn't give an origin of the construct state. I considered maybe the construct state might be derived from a 3rd person possessed marking which might in turn be derived from a pronoun or a verbal person marker, but none of the languages that have this EA morpheme also have any pronouns/person markers with /n/ at all.

Also, assuming it was originally a separate word that cliticized onto the stem... why does it occur before the case ending *-om? I guess it's just so ancient it preceded even the development of case markers? Or did it metathesize into the stem, and if so, why, since *-om-Vn is not particularly hard to pronounce?

Answer 1

How about a topic marker, like Japanese wa? You could start with a deictic of some sort ("this thing right here"), which got semantically bleached into a general marker of new information ("this thing is important" > "this is the core of the clause"), and then got bleached even further into a head marker ("this is the core of the phrase").