Reconstructions

PC
*dagan- - grain (Kogan 2015: 288, Renfroe 1992: 91‒94, del Olmo Lete 2001, Schwemer 2001: 282)
Attestations of *dagan- in Aramaic are sporadic and may be due to Canaanite influence
PC
*gagg- - roof (Greenfield 1969: 98, Ginsberg 1970: 103, Kogan 2015: 289)
PC
*gin-t- - wine or olive press (Kogan 2015: 292)
For probable connection to *gann-at- - ‘garden’ see Michaut-Colombot 1997 and Heltzer 1999:196‒197 Arb. wǯn ‘to beat (usually about clothes beaten by a fuller)’ is semantically too remote to be a reliable cognate (cf. BDB 387, HALOT 206, DRS 493‒494)
PC
*grš - to cast out, to expel (Kogan 2015: 289)
. The hypothetic Phoenician attestation of this root (ngršin KAI46:2) is not universally accepted (cf. DNWSI 236, Krahmalkov 2000:144)
PC
*gašm- - rain (Kogan 2015: 291)
+ Eg.-syll. gas-mu ‘storm’ (Hoch 1994:354), not universally accepted, cf. Rainey 1998:450, Woodhouse 2003:281
PC
*gyl - to rejoice (Kogan 2015: 289)
probably derived from PWS *gwl/*gyl ‘to move in circle, to turn, to dance’
PC
*γār- - (loose) skin, dewlap (SED I No. 106; Kogan 2011: 216)
has no etymology
PC
*hlm - to hit, to strike (Kogan 2015: 292)
Cf. perhaps (with Leslau 1958:18 and DRS 417) Tna. halämä ‘to smack someone in the face, to box his ears’
PC
*hm - to be noisy (DRS 419)
biconsonantal element
PC
*hamull-at- - crowd (DRS 419, Kogan 2015: 292)
is probably an extension of the biconsonantal element *hm ‘to be noisy’ Arb. hml ‘to overflow and pour forth (water, rain)’ compared in HALOT 251 is semantically remote; Arb hamūlat- - ‘(a herd) left to graze by itself’ is also not connected to the present root