Two Cases of a Pseudo-Dual in Biblical Hebrew
I’m currently at the Summer School in Languages and Linguistics at the University of Leiden in The Netherlands. (N.b., Leiden is perhaps my favorite city in all of Europe, mostly for historical reasons!) I’m taking four courses: Northwest Semitics (Ugaritic and Phoenician), Advanced Biblical Hebrew, Sabaic, and Historical Grammar of Hittite. It was interesting that one particular topic of conversation was brought up today, independently, in three of the four classes, namely the morphology and semantics of the Hebrew word for “sky, heaven” (שָׁמַיִם šāmáyim; pausal שָׁמָ֑יִם). According to HALOT, the word occurs 420 times, while Accordance gives 421 hits for a search of the word. The interesting thing is that it appears morphologically like a dual (cp. the forms רַגְלַיִם ragláyim ‘two-feet’ and עֵינַיִם ˁênáyim ‘two-eyes’). Of course, the dual was actively used in earlier phases of Northwest Semitic, as evidenced by Ugaritic; however, by Iron Age Hebrew it survives mostly in naturally occurring pairs, mainly body parts. Not including the words שָׁמַיִם and מַיִם, an Accordance search of all dual nouns in biblical Hebrew yields 2,013 results; a quick scan of the examples in Genesis shows these examples: nostrils, eyes, two years (שְׁנָתַיִם in Gen 11:10; 41:1; 45:6), feet, palms, ears, arms/hands, two times/twice (פַעֲמַיִם in Gen 27:36; 41:32; 43:10), knees, loins, two teeth (שִׁנַּיִם in Gen 49:12), two sheepfolds/saddlebaskets (?) (מִּשְׁפְּתָֽיִם in Gen 49:14), and breasts. Two observations are immediately obvious: (1) body parts that come in pairs clearly dominate the list, and (2) out of the four other examples (two years, two times, two teeth, two mišpətāyim [?]), two of them come from the archaic poem in Gen 49 which, theoretically, represents an older period of the language when the dual was still operative. The purpose of this blog post is to float the idea that one particular case (and possibly others as well) of what appears to be a dual is what I call a pseudo-dual, namely the word for “sky, heaven” (שָׁמַיִם). My thesis is that this form is best understood, historically, as a plural form of the III-y word *samay- which was written defectively according to Iron Age orthographic conventions, only to later undergo an analogy with the duals that shifted the stress and thereby was (synchronically) reanalyzed as a dual in the Masoretic reading tradition(s). In fact, the argument to analyze these forms as plural rather than dual is not original to me, but dates all the way to Paul Haupt (the first professor of Semitics at Johns Hopkins!) in 1893, and can be found more recently in Gesenius, Kautzsch, and Cowley’s Gesenius’ Hebrew Grammar §88d. However, my explanation of how the historical plural came to be understood synchronically as a dual via analogy and reanalysis (see below) is, to my knowledge, original to me. But only a thorough review of the grammars and literature (hopefully forthcoming) will confirm whether this line of reasoning has been previously suggested.
The essential points of my argument are as follows:
(1) The reconstruction of the word for “sky, heaven” in Proto-Semitic is *samay- with a III-y, despite the Arabic plural form samawāt with III-w. That is to say, Ethiosemitic samāy and NWS *šamay– necessitate Proto-West-Semitic *samay- (Akkadian šamû is of no further assistance in the reconstruction). {Without having reread Suchard and Groen 2021 recently, is it possible that the /w/ of Arab. samawāt is related to the form of the *CVCaC-ū- plural which ultimately derives from a pre-Proto-Semitic plural suffix *-w– which underlies the external plural morphemes? Prima facie, it seems difficult to suggest this since their argument rests on the analysis of the noun form being *CVCC- which, when adding a plural suffix *-w-, results in a consonant cluster that is resolved by an epenthetic *a.}
(2) Pre-Masoretic Hebrew orthographic conventions were that scribes did not write matres lectionis when the preceding consonant of the mater was already of the same quality. For example, the plural word for “nations” is most often written as גּוֹיִם (136x; not **גויים) compared to just a single odd spelling of גֹיִים (Gen 25:23 which is otherwise corrected in the qere to גוֹיִם). However, note that this is not the case when two adjacent consonants of the same quality are both consonantal, as in יַיִן yayin ‘wine’ (cp. the form ין yên in the Samaria Ostraca which not only evinces monophothongization of the diphthong in the Israelian dialect, but also the aforementioned principle that the historical spelling of יין for yên would violate Iron Age orthographic conventions). One apparent exception to the orthographic principle above is the plural word חַיִּים where the second yoḏ must be understood as a mater.
(3) Comparative Semitic evidence also suggests that we should expect the plural instead of a dual form which would otherwise appear odd. Among the other Semitic languages, Akkadian uses the plural šamû as well as Ugaritic (written consonantly both as šmm and šmym, but syllabic transcriptions show ša-mu-ma of the nominative plural).
(4) One obvious point to deal with is the vowel pattern and penultimate stress (šāmáyim, not **šəmayím) which makes it appear to be a dual rather than plural (i.e., because plural forms are accented on the ultima with propretonic vowel reduction, e.g., məlāḵī́m). This is why analogy and reanalysis is a necessary step in the process. Before presenting my analogy, however, it is worth noting that one finds two previous suggestions in GKC §88d: “Only apparently dual-forms (but really plural) are the words מַ֫יִם water and שָׁמַ֫יִם heaven. According to P. Haupt in SBOT. (critical notes on Isaiah, p. 157, line 18 ff.), they are to be derived from the old plural forms (found in Assyrian) mâmi, šamâmi, whence the Hebr. מים ,שמים arose by inversion of the i mâmi, mâimi, maim. It is simpler, however, to suppose that the primitive singulars may and šamay, when they took the plural of extension (§ 124 b), kept the tone on the ay, thus causing the îm (which otherwise always has the tone, § 87 a) to be shortened to im” (GKC, 1910: 246). While the explanation by GKC is preferable to Haupt, it’s ad hoc to suggest that the plural of šamay- retained what would have been at the time pre-penultimate stress, i.e., šamáyīma. Instead, it’s even simpler to say that the shift of accent in the form šāmáyim arose on analogy with the similarly sounding dual forms and was later reanalyzed as a dual, i.e., šāmáyim << *šəmayī́m < *šamayī́ma. This process was aided by the already defectively spelled plural שׁמים with one yoḏ instead of two, which appears, prima facie, as a dual anyway.
In conclusion, one final note is in order: the astute observer will notice that I titled this blog post as “Two Cases of a Pseudo-Dual in Biblical Hebrew” but only discussed one (שָׁמַיִם). The second case which probably reflects a similar situation is the word for “water” (מַיִם máyim << *mayī́ma; see already the discussion about Haupt and GKC above). Taken together, the forms of שָׁמַיִם and מַיִם are what I call pseudo-duals in biblical Hebrew, at least from the diachronic and comparative perspectives.
Postscript
I want to extend my gratitude to Agustinus G. and Benjamin S. for their inspiring teaching, enlightening publications, and insightful conversations, through which I’ve learned a great deal about historical Hebrew grammar. Whether or not they would find my reconstruction of the forms שָׁמַ֫יִם and מַ֫יִם convincing, however, I do not know.
About The Author
Rare Semitic Languages 
Semitic Consonant Correspondences