The Proto-Indo-European word *mori is one of the many words that have been reconstructed for the language without much debate. It is present in enough of the daughter languages, and follows the expected sound change laws that we would expect, and so may be confidently added to the dictionaries and word lists. The only problem with the word is that its meaning is precise but not exact. In most of the daughter languages it means either ‘sea’ or something very similar to it. It may not have always meant sea however, but rather may have referred to a body of water of some type, or more broadly to any type of body of water, such as river, swamp, lake, or ocean.
Its reconstruction with the meaning of ‘sea’ is based on inter alia, the following words found in Indo-European daughter languages:
Proto-Balto-Slavic: *mári
– Lithuanian: mãrios
– Russian: móre
Proto-Celtic: *mori
– Irish: muir
Proto-Germanic: *mari
– German: meer
– English: mere
Proto-Indo-Iranian: *maryáHdaH
– Sanskrit: maryā́dā
– Ossetian: mal
Proto-Italic: *mari
– Latin: mare
Old Armenian: mawr
In the Hittite language, we find that the meaning is somewhat different. Rather than ‘sea’, we have this form:
Hittite: marmara ‘swamp, marshland’
Perhaps this is not surprising, as the Hittite homeland was located in central Anatolia, a land quite far from any sea, and containing many marshes. If the Proto-Indo-European speakers migrated into Anatolia via an inland route, then they would not have been exposed to the sea and so may not have a word for it. It has been argued therefore that the PIE word *mori originally referred to something like a swamp or a marsh, and then later came to mean sea in late-PIE (ie. post-Anatolian and post-Tarim), from which all of the other daughter languages inherited the meaning ‘sea’. What is not in doubt is that the original PIE term *mori referred to a body of water of some type.
It may be speculated that the word *mori looks somewhat peculiar, appearing in the o-grade and ending in -i-, and that the word must come from a root *mer. I believe that this is likely, as it fits the pattern of almost all other PIE words which can be traced back to a root type *CeC-. Nonetheless this is difficult to prove, as no other words which could derive from this root may be found, so we must be content to reconstruct *mori and to only speculate about its PIE etymology.
Unless of course, we can find another source for the word.
The Proto-Afroasiatic root *m·r
The strongest candidate by far is the PAA root *m·r which also refers to various bodies of water. It is reconstructed based upon the following words from the Afroasiatic daughter languages:
Egyptian: m·r ‘canal, pool, waterway’
South Omotic *mir- (“river”)
Aari: mɨri ‘river, stream’
Dime mɪ́rɛ river’
Proto-Cushitic: *mVr
Oromo: meerii ‘watering trough’
Tsamai: mīre ‘pond’
Proto-Chadic: *mVr
Somrai: mā̀rī ‘rain’
Fali: mirə̂ ‘river’
Muskum: mìrà ‘oxbow lake, tributary stream’
There may also be a Semitic source available, however it is not as clear. There is a Sabaean word m·r which is related to agriculture and irrigation, however its exact meaning is unclear. So too there is a Hebrew word mar which means ‘drop of water’, although it is only attested in a single verse of the Hebrew Bible. Based on the limited evidence, a Proto-Semitic form would likely be reconstructed as *mar- and would have a meaning related to water.
Similarly there seems to be a Berber form, based on words such as Kabyle əmmir ‘pour’, derived from a Proto-Berber root *mVr.
The reconstruction of a Proto-Afroasiatic word based upon the attested forms in the daughter languages is possible, and it should be considered to be strong as forms are present in both southern and northern Afroasiatic branches. If we include Semitic and Berber (ařimam ‘lake’ may be asserted if metathesis occurred) then the word is found in every branch of Afroasiatic. Even if we exclude Semitic and Berber, it is still able to be reconstructed with confidence, and this is why it regularly appears in Proto-Afroasiatic wordlists.
Ehret reconstructs the PAA form as *mur ‘to flow’
Stolbova reconstructs the PAA form as *mar ‘drop, rain’
I have not found any reconstruction by Militarev, nor any other well-known sources, although Wiktionary offers A PAA form *m-r meaning ‘river, channel’ without any sources indicating the reasoning behind the reconstruction.
The reconstruction of a Proto-Afroasiatic form *m·r is reliable, and it may be given a meaning related to water or flowing of water. I believe that this helps to establish that the Proto-Indo-European form may also go back to a root form of *mer, and that it is inherited from the Proto-Afroasiatic root, sharing both the same consonantal structure and general meaning.
For reasoning why I believe that the Proto-Indo-European language is in fact a member of the Afroasiatic language family, and is most closely related to Semitic and Egyptian, you may read the following pieces:
100 Proto-Indo-European Words that are inherited from Proto-Afroasiatic
The Afroasiatic Hypothesis (the Origin of Proto-Indo-European)
Linguistics and Nonsense 