In most environments in Alb., PIE palatals eventually became interdental th and dh (opt. > d); in Iranian languages like Old Persian, the same interdental th and dh (all > d) occurred but in different environments. In still different environments in Arm., if PIE palatals eventually became interdental th and dh > d as well, with further changes later, it would help explain some odd correspondences.
First, in:
ager -ro- = field L; agrós G; ájra- S; art -o- = tilled field Arm;
a simple gYr > dzY > zYr > zr > Dr > dr would explain all (with later met. of dr > rd as in dákru G; artawsr Arm; etc.). This was most likely to happen if the new palatals were dental unlike other fricatives (that is, if at the time PIE s first changed to retro. s. > s^ in some environments, and s > ? > x > h in different environments (more on s>h to come later)).
Second, there were further changes of th > f and then all other changes as in p > f ( > h/0/w/etc. ) for kY before r (also tr > Tr to explain -tr- > -wr-, merging just as dr/ gYr above, more to come later). This also happened to secondary kY (from gYH and gY, kY and skY before voiceless C (needed to devoice gY()). This would explain all these correspondences:
mjekër = beard Alb; smakrà Lth; mawrukH Arm; çmáçru- = beard/mustache S;
know E; c^anac^Hem; canawtH -i- = an acquaintance Arm;
*slaxskYè+ > aLac^Hem = pray; aLawtH-kH (pl) -i- = prayer(s)/supplication Arm;
*sislaxskYè+ > i:láskomai = appease gods G;
aracem (trans) = pasture; arawt -i- = pastureland Arm;
in
tró:go: = gnaw/nibble G; tresk- = chew Toc; aracim (intr) = browse/graze; aracem (trans) = pasture; arawt -i- = pastureland Arm;
októ: G; owtH Arm;
So KYt merging with Tt (t+t or d+t), including tt / kYt just as tr / kYr and dr/ gYr above, more to come later.
A little explan. is needed for a few others:
*weikY+ > vi:cus = village L;
*wikY+ > vIsI OCS; Gis = (a village in E) Arm;
*weikYla:x \ -i:x > vi:lla = country-house/villa L; *weywli > giwL \ gewL \ geL (nom) geLj^ (gen/d) gewLkH (pl) giwLicH (gd) = village Arm;
if kYl also > sYl / sl with opt. dis. of w-w > w-0 at the right stage; yl > ly > ljY, etc.
*plekYtí+ > hiws -i- = plait, (*deriv. v) hiwsem = weave/plait Arm;
with opt. pl > xWl > h(l) / w(l) / xW etc. (seen in modern dia. with evidence of xWl in l- or f- or x-)
also seen in
*plew+ > hosem = make flow / pour down Ar;
*plew+ > hiwsi \ hosi(n) = avalanche Ar; >>
husi Xot; husin Bul; fOsan Arar;
and opt. kYt > kYs as in
seca:re = cut off, sectus, *sectus > sexus (m) -u:s (g) = *division > sex [i.e. male/female] L;
among other similar widespread changes. The common ending -tu- / -ti- is seen often in these ex.
The changes of cluster kYs > cYs. in:
6; vecH Arm;
16; ves^tasan Arm;
60; vatHsun Arm;
with -cYs. > cY but > s. before t and s.>0 before s (as sYsY > ssY > TsY > Ts > tHs) show new sYsY or simil. underwent > T but became a stop before T > f > w. The compounds could also have been made at dif. times, so no certain original ev. comes from this.
The old analogy posited for owtH getting -w- from ewtHn is not needed, as seen. The changes are likely opt. (at the stage sYC > sC, with s > T but sY unchanged at the time, etc.) due to similar environments with -s-, such as dustr. It is not known if this happened before met. in:
dákru\dákru:ma G; artawsr (n) artasukH (pl) Arm; tear
though either way would be fine if also opt.
Though other non-opt. ways to explain some of these EXTANT forms exist, all of them together, as well as the number of other opt. changes in Arm. suggests the analysis here.
PIE palatals in Arm. (kYl- with pl-)
In most environments in Arm., PIE palatal kY eventually became s, but kYl- has a disputed outcome, some say > sr- but more often l- (merging with pl-) in secure ex. This makes most sense if opt. kY > sY > s > th > f as p > f etc, as in previous ex.
First, in:
lu = heard / news / fame Ar; klutós = famous G; çrutá- S; inclutus L;
lur = hearing / news / voice Ar;
lsem Ar; slys^ati OCS; klyaus.- TB; klausytun OPr;
clue:re = hear L;
kYl- > l- in secure ex. for hear.
Second, there were further disputed changes for slope / incline:
linim = be Ar; klí:no: = make slope / incline / lean G;
leaRn = mountain Ar;
hlains = grave/hill, hlaiw = grave, hlaiwasna = grave Got; cli:vus = slope/declivity L;
hlíð = slope/mountainside ON;
hlinc = ridge / bank / hill OE;
This matches the common outcome of pl-:
lu Ar; plús.i- S; louse
luanam = wash Ar; plúno: G;
lnum Ar; pímple:mi G;
li Ar; pra:tá- S;
plé:re:s = full G;
against disputed changes for pl > pHl (which could easily be opt. as p- > h / pH ).
Also, as suggested by others, to match my gYr > rt there might also be rgY > rt :
birce OE; birch E; bérz^as Lth;
barti = poplar Ar; (as Alb. rdh etc.)
brad = white fir Ruman;
though this would require restructuring and there are other expl.