Edenics

Edenics is a project by Isaac Mozeson to trace all languages to Edenic, his term for the early stage of the Hebrew language that he claims was the common language of humankind spoken before the confusion of tongues at Babel.

Identification of Proto-World with Hebrew
In the real world, Hebrew is a language of the Northwest Semitic branch of the Afroasiatic family. It had been traditionally identified as the Proto-World language prior to scientific linguistics. For example, Dante Alighieri concluded in chapter "Paradiso XXVI" of his Divine Comedy that Hebrew was derived from Adam's language. The article about the Hebrew language in Insight on the Scriptures, a Bible encyclopedia published by Jehovah's Witnesses, states that proto-Hebrew was miraculously less affected by the confusion of tongues at Babel than other languages. Mozeson agrees and is selling a $15 PDF dictionary titled E-Word that lists thousands of alleged cognates between Hebrew and other languages.

Mozeson is by no means the first to set forth detailed correspondences between Hebrew and European languages. In 1869, Robert Govett wrote a book listing his own Hebrew false cognates. In the first two pages, he followed Adam's example in naming the animals: "rabbit" from arn e v e t, "snake" from n a kh a sh (with S mobile prefix), "asp" from tz e f a ` "viper" via Greek aspis, "lion" from l a vi '  (with weakening of v to an approximant), "steer" from shor "ox" (blended with its Aramaic counterpart toor[!wikt]), badger from a metathesis of ` a kb a r "mouse", and "goat" and "kid" from g i di. A "kitten" is a q i t o n "little one", and a "giraffe" has a long ` a r a f[!wikt] "neck". He explained elephant as al e f[!wikt] "draft animal" combined with a suffix -ant presumably meaning "-like", reversed them for "antelope", and dropped the suffix for Greek elafi "deer". His camel from g a m a l is accepted, and jerboa from tz a v o ` a  "hyena" appears plausible given the accepted Arabic derivation. But his "adder" from ` a t a r[!wikt] meaning "encircle" is a false cognate to the accepted etymology (akin to Latin natrix with the initial n reinterpreted as part of an), as are his "boa" and "viper" from af o `[!wikt] ("viper" comes from Latin meaning "viviparous"). But with this many coincidences among land animals alone, even before he gets to birds, one begins to wonder whether it actually is coincidence or in fact an echo of Semitic phonosemantics in western Indo-European.

Dr. Zaidan Ali Jassem has done similar analysis linking Indo-European words to Arabic, another Semitic language. Other linguists such as Grzegorz Jagodziński at least admit that many IE-to-Semitic correspondences are false cognates, though Theo Vennemann and others have hypothesized Germanic loans from Semitic. .

Equivalence
In E-Word, Mozeson groups the letters of the Hebrew alphabet into seven equivalence classes whose members share a place or manner of articulation. These equivalence classes incidentally are reminiscent of those used in Soundex indexing, and there are as many of them as there are syllables in Solresol. He defines a color for each in the glosses:

Mozeson claims that reflexes of an Edenic root in a child language may be subject to arbitrary shifts (replacement of a consonant with another of the same class), nasalization (insertion of N or M between consonants, such as the Proto-Indo-European nasal infix), metathesis (exchange of adjacent consonants), or even complete reversal of a triconsonantal root. The last of these resembles the reverse speech pseudoscience of David John Oates (no relation I assume).

Validity
Except one would have a hard time distinguishing Mozeson's results from chance. Allowing shifts within an equivalence class, there are P(7,2) = 7×6 = 42 roots with two distinct letters and P(7,3) = 7×6×5 = 210 three-letter roots. But in families other than Afroasiatic and Indo-European, Mozeson is liberal with metathesis and reversal, taking order out of the picture. This leaves only C(7,2) = 42/(2×1) = 21 distinct two-letter roots and C(7,3) = 210/(3×2×1) = 35 with three letters, a smaller vocabulary than even Toki Pona and fertile ground for coincidence. Overall, Mozeson's methods resemble the exploratory technique of multilateral comparison popularized by Greenberg and Ruhlen, which has largely been rejected by linguists as finding more false cognates than provable descents. Mainstream linguists would probably call Mozeson a crackpot.

But the false cognates could still be used for a constructed language in speculative fiction as an attractor in the dynamical system of language change, in the same way that similarity in unrelated languages' words for mother, father, and food is a consequence of how babies learn language. Especially in a conworld with an Abrahamic creation myth, this might be explained through latent cross-linguistic phonosemantic associations that Hebrew exploited but which were weakened after the Babel incident. Such a conlang will be lexically about as close to Hebrew as Earth languages are.

And if anything, it's a fun way to learn Hebrew, just as Chinese-English false cognates help English speakers learn Chinese.