From inflexion to derivation: The PIE word for 'salt'
The Proto-Indo-European word for 'salt' has reflexes in all the main branches of the Indo-European languages with the exception of Anatolian. It can be reconstructed as a neuter l-stem: *séh₂-l, *sh₂-él-.
A unique feature of this word is a stem-final element *d in some but not all branches of the family. Evidence for *d is found in Armenian, Germanic, Italic, and Balto-Slavic, with a possible further trace in Indo-Iranian. Since *-d- cannot be explained as an independent innovation that occurred in each branch, this formant was most likely an integral part of the late Proto-Indo-European paradigm of the word for 'salt'.
In this paper, I propose that Late PIE *sh₂éld continues the Indo-Anatolian instrumental singular form *sh₂élt which came to be used as a subject marker when the protolanguage switched from ergative to nominative alignment. The attested wordfinal *-d is the result of a phonetic shift *-t > *-d that happened after the Anatolian languages had split off. The final *-d of *sʕald was lost in Tocharian and Greek, and was reanalyzed as part of the stem in other branches.