Dear Caley,
Thank you very much for your critical feedback; I appreciate it very much.
Regarding the semantic development of ari-, I appreciate and I am grateful for Caley's observations about the etymology and historical semantic development of this form in Indo-Aryan. I am curious, does this mean that you claim that Monier-Williams was mistaken in his definition of 2. a-rí- as "'not liberal,' envious, hostile, RV.; (ís), m. an enemy, RV." (M-W p. 87, bottom of 3rd column, long final entry in the column)? My interpretation of this entry would be that according to Monier-Williams, this word appears in the Rigveda (RV) with the meaning "an enemy". Does this mean that MW's interpretation of the relevant passage of the Rigveda was incorrect? Changing the meaning of a word from "enemy" to "guest" seems to be a significantly major alteration that would drastically change the meaning of the passage of RV in which it appears.
But if we accept that MW is indeed mistaken on this semantic point, as Caley suggests that Mayrhofer indicates, then we may still return to the presumably original meaning "other" or "other person". It still seems plausible to me that there may well possibly have been an independent semantic development in the Mediterranean in the Bronze Age that could have been parallel to the later post-Vedic in situ semantic development: "other (person)" > "enemy" seems to be a natural enough semantic development that could have occurred independently in different times and places from the same Indo-Aryan root word.
As a typological comparison, we may consider the semantic development of Proto-Indo-European *gʰóstis, which developed to mean hostis 'enemy' in Latin, but which developed to mean gasts 'guest' in Gothic, gestr 'guest' in Old Norse (from which indeed English "guest" is derived), гость 'guest' in Old Church Slavic, etc. Likewise we may consider the semantic development of Polish obcy 'foreign; stranger' and dialectal Ukrainian ві́бчий 'foreign', both from Proto-Slavic *obьťь 'common', a meaning retained in Old Church Slavic and other Slavic daughter languages. Another example is Proto-Slavic *ťȗďь 'foreign, alien, strange' (e.g., OCS щоуждь, Russian чужой, чуждый) from PIE *tewtéh₂ 'people, tribe'; Baltic cognates largely retain the original PIE meaning or develop it to mean 'land, country', but one extended meaning of Latvian tauta is 'people from another region'. (The PIE root is the ancestor of German Deutsch, Proto-Italic *toutā, Irish tuath, Welsh tud, etc.)
The point is that numerous typological examples demonstrate that the meaning 'enemy', 'hostile', 'foreign', etc., can frequently arise independently from a process of historical semantic development from roots with more neutral original meanings. Thus it seems plausible to me to suggest that the semantic development 'other (person)' > 'enemy' and/or 'other (person)' > 'guest' > 'enemy' could have occurred independently in Minoan Indo-Aryan at an early stage and in post-Vedic Sanskrit at a later stage.
Best,
Geoffrey