*(revised after the clarification given in the initial comment)*
Genesis 1.1 is one of the most discussed verses in the Hebrew Bible. It is the first verse of the first book, initiates the Hebrews’ grand cosmology, and … contains an apparent grammatical crux. Phooey! You would think that one could get further than one word into the Hebrew Bible without a grammatical problem.
In fact, there is no problem, only a long-term misunderstanding of Hebrew grammar. In a 2008 article appearing in Vetus Testamentum (which revised a sub-section taken from my 2002 thesis), I argued for an analysis of the first verse that is grounded both in my long-term research on the Hebrew relative clause and comparative Semitic grammar. You can find the article linked here.
But recently I was criticized (on a blog), for failing to explain how my analysis of 1.1 fit into an interpretation of 1.1-3. So, although my argument for Gen 1.1 stands ably on its own, I will take the opportunity presented by the recent criticism to summarize my argument for 1.1 and provide my analysis of vv. 1-3.
The Nature of בראשׁית in Gen 1.1
In a nutshell, the interpretation and translation of the first complex word, בְּרֵאשִׁית, in the Masoretic text of the Leningrad Codex as an absolute temporal prepositional phrase, “in the beginning, …” is grammatically indefensible. Period. End of story.
If one wants to ignore the Masoretic vocalization and read the word with an articular vowel with the preposition, i.e., *בָּרֵאשִׁית, “in THE beginning,” as the Samaritan Pentateuch appears to do, fine. But one must not only recognize that such a choice is a departure from the Masoretic text, but also fails to explain the Greek Ἐν ἀρχῇ, which also lack the definite article.
What is the grammatically justified analysis? The noun ראשׁית is bound to an unmarked relative clause, “beginning-of (that/when) God created …”. This construction, which is found in Ge’ez, Old South Arabian, and Akkadian, must be as old as Semitic itself. In other words, the noun-bound-to-clause structure of ראשׁית ברא in Gen 1.1. finds a clear parallel in the Akkadian pattern di:n idi:nu “judgment (that) he judged/rendered” (Lipinski 2001:533-34; also see Deutscher 2001, 2002 for insightful linguistic discussion of origins of the Old Akkadian relative clause).
Here I should also mention the excellent study, Baasten 2007. Baasten covers much the same ground as I do in my 2008 VT article and it is unfortunate that our library did not receive the book until well after my article came out. I recommend reading Baasten’s study alongside mine.
The biggest difference between Baasten’s study, as well as all previous studies of the noun-bound-to-clause construction in Semitic, and my argument (in my thesis and in the 2008 VT article) concerns the semantics of this unmarked relative clause. I argue that using a bound form of the noun serving as the head of the relative clause is one strategy used to mark the relative clause as restrictive. The other strategy used to mark a Hebrew relative as restrictive is to omit the relative word, i.e., an unmarked or asyndetic relative clause. Interestingly, both strategies are used in בראשׁית ברא! That is, Gen 1.1 is doubly-marked as a restrictive relative clause, meaning that this particular ראשׁית cannot be identified without the information given within the relative. It is the particular ראשׁית during which God created the heavens and the earth. It is not an absolute ראשׁית, “THE beginning”, but just one specific ראשׁית that is being referenced in Gen 1.1.
That is the essence of my argument for Gen 1.1.
The Addition of the Preposition ב to בראשׂית
The difference between the basic di:n idi:nu syntax of ראשׁית ברא … and how Gen 1.1. really starts is the addition of the clitic preposition ב. The preposition takes ראשׁית, with its attached relative clause, as its own complement. The lack of the articular vocalization in the Masoretic tradition leaves open the question whether the ראשׁית should be translated as definite in English. Since ראשׁית is “in construct,” it depends on the definiteness of its clitic host to signal its own definiteness. The problem, of course, is that a clause is never marked as definite. So, we are left with some ambiguity: is בראשׁית ברא “in a beginning period that …” or “in the beginning period that …”? I suggest that the referential nature of the nominalized clause grounds the ראשׁית sufficiently to make it definite (specific, identifiable) and so using “the” in English, as long as no comma is inserted after “beginning,” is the legitimate translation.
As a whole, the addition of the ב preposition indicates that the noun ראשׁית and its relative clause have been assigned a role within a larger clause. Fronting a prepositional phrase as a scene-setting Topic before the main verb is a very common narrative strategy in Hebrew. Once need only look for examples like Gen 22.4, בַּיּוֹם הַשְׁלִישִׁי וַיִּשָּׂא אַבְרָהָם … “on the third day Abraham lifted …”. Moreover, once it is recognized that
the vast majority a high percentage** of occurrences of וַיְהִי in BH narrative are discourse markers and not the main verb for a following prepositional phrase, the use of Topic PPs fronted before a wayyiqtol (past narrative) verb becomes ubiquitous at scene transitions in narrative. In other words, in Gen 4.3, וַיְהִי מִקֵּץ יָמִים וַיָּבֵא קַיִן…, the initial ויהי is a discourse marker and the PP מקץ ימים is a Topic-fronted temporal modifier for the verb ויבא, “At the end of (so many) days, Cain brought …”.
**According to the study in this new post, the discourse ויהי account for 48.4% of the total ויהי in the book of Genesis.
The Status of Verse 2
The Topic-fronted PP, main wayyiqtol verb pattern of Gen 1.1. is very well-attested in biblical narrative. So what role does verse 2 play, with its shift to a Subject-Verb (qatal/perfective) syntax? The simple answer is that it is a compound parenthesis, consisting of 3 clauses.
Parentheses are constituents (phrases, clauses, or even compound clauses, like Gen 1.2) that interrupt the flow of an ‘argument’, whether the argument is at its core chronological (i.e., a narrative) or logical (i.e., an exposition, as in, e.g., many psalms).
The easiest parenthetical constituents to identify are those that are syntactic interruptions, as in Esth 9.24, וְהִפִּיל פּוּר הוּא הַגּוֹרָל לְהֻמָּם וּלְאַבְּדָם “and one cast a ‘pur’ (it is the lot) to disturb them and destroy them.” In Esth 9.24, the null copula clause הוא הגורל “it is the lot” interrupts the clause within which it sits, separating the core of the main predicate from the adjunct infinitive clauses. Note, though, that parentheses cannot simply be thrown anywhere in its host clause. Rather, they must be placed at word or phrase edges. In other words, one never finds a parenthesis that intervenes between a preposition and its complement, since those two items either form a word (i.e., when the preposition is ב, כ, or ל) or a phrase in which one or both parts cannot stand on their own (i.e., even the preposition is orthographically separate, it still ‘leans’, i.e,. is cliticized, on its complement host). This also applies to collocations of verbs and complements. So, in the case of Esth 9.24, the parenthesis is inserted between the verbal complement and the verbal adjunct. We would not find a parenthesis intervening between the verb and its complement, because those two items combine to form a semantic unit.
A test for parenthesis is to ask these questions:
1. Does the clause in question add an event on par with the preceding event? If so, it is not likely a parenthesis.
2. Does the clause in question add information about a specific constituent in the preceding clause and yet does not appear to be a relative clause? Also, does the clause in question overlap with the preceding clause in almost all the constituents but adds, say, one new constituent? If either (or both) is true, the clause is likely a parenthesis. (Note that this condition distinguishes parenthesis from what are often taken as circumstantial clauses, where the overlap is minimal and the clause does not primarily modify a single constituent, but often an entire event or situation).
3. Does the structure of the clause in question differ from the structure of the clause on either side of it and do those two clauses share a similar structure? If so, and if it does not contribute an action or even on par with the preceding and following clauses (per #1), it may be a parenthesis.
With this description of and criteria for identifying parentheses in hand, we can now turn back to Gen 1.2. Syntactically, the compound clause in v. 2 sits between a Topic-fronted adjunct clause, בראשׁית…, and the main verb, ויאמר in v. 3. So far, so good—it sits at an appropriate phrase edge. Does it add an event on par with the preceding event (#1)? In the case of Gen 1.1-3, this criteria is hardly applicable, since the preceding event is also subordinate. But criteria #2 fits perfectly, since Gen 1.2 picks up with הארץ, which was first presented in v. 1, but then adds something more. So there is overlap, but also additional information. And finally, #3 seals the identification: there is clear structural difference in syntax between, on the one hand, v. 2, with its Subject-Verb order and, on the other hand, the noun-bound-to-clause in v. 1 and the wayyiqtol clause in v. 3. Now, v. 1 and v. 3 don’t share the same structure, but that’s because v. 1 a syntactic part of v. 3. Nonetheless, the shift we see in v. 2 is paralleled many times with other parentheses in the Hebrew Bible.
The overall analysis of Gen 1.1-3 given above has a long history in biblical scholarship. It is also the analysis adopted in Baasten 2007, although with the tie-in to relative clause restrictivenes. Here is a basic English translation that would serve as a starting point for working out one that reflects whatever translation theory one adopts:
“In the beginning period that God created the heavens and earth (the earth was formless and void, and darkness was over the surface of the deep, and the wind of God was hovering over the surface of the waters), God said, ‘Let light be!’”
Happily, within biblical scholarship, the analysis I have promoted above is being adopted by my peers (e.g., Mark Smith in his Priestly Vision of Genesis 1, and Ellen van Wolde, in her 2009 JSOT article). [John Walton's book was submitted in 2006, too early to have read either Baasten's or my articles.]
More troubling is the attitude exhibited in the critical blog I mentioned at the outset. The owner [RDH: the author requests attribution], Peter Kirk (gentlewisdom.org)
presents himself as a representative of was formerly a professional Bible translator but is no longer; he wrote this:
But the traditional understanding of 1:1 has a history going back over 2000 years to LXX. To overturn such a tradition you will always need overwhelming evidence. And neither I nor the majority of professional Bible translators have seen that overwhelming evidence. So for the moment you need to accept that your position is considered one of the possible alternatives …
Beside the mistaken view of the LXX on Gen 1.1 (it does not reflect an article, just like the Masoretic vocalization; see above), what is disturbing about this comment is that professional Bible translators, to my knowledge, rarely have PhDs in biblical studies and are thus not experts in Hebrew grammar or Hebrew exegesis. [RDH: Peter Kirk has clarified what he referred to by professional translators—those who made English translations like the NIV11; this was not how I was taking him; see my comment below.] And yet, this one has clearly set himself up as a greater expert on the grammar, textual tradition, and literary features of Gen 1.1 than me, Martin Baasten, Mark Smith, Ellen van Wolde, and over a century of Semitists who understood the basic grammatical noun-bound-to-clause structure of Gen 1.1.
Disturbing, indeed, and not a good sign for the quality of the interpretation behind Bible translations done by “professionals” with this relationship to Hebrew grammarians.
[RDH: clarification in the comment below. I do not mean to disparage all professional translators, of both types, as I specify in my comment; rather, their respective skill sets indicate that one would think both would look to developments in Hebrew grammatical analysis. Hmm...digging self deeper hole. Oh well.]
* Baasten, Martin F. J. 2007. First Things First: The Syntax of Gen 1:1-3 Revisited. Pp. 169-88 in Studies in Hebrew Literature and Jewish Culture Presented to Albert Van Der Heide on the Occasion of His Sixty-Fifth Birthday, ed. Martin F. J. Baasten, and Reinier Munk. Dordrecht, the Netherlands: Springer.
* Deutscher, Guy. 2001. The Rise and Fall of a Rogue Relative Construction. Studies in Language 25 (3):405-22.
* Deutscher, Guy. 2002. The Akkadian Relative Clauses in Cross-Linguistic Perspective. Zeitschrift für Assyriologie 92:86-105.
* Lipiński, Edward. 2001. Semitic Languages: Outline of a Comparative Grammar. 2nd ed. Orientalia Lovaniensia Analecta 80. Leuven: Peeters Publishers.
* Smith, Mark. S. 2010. The Priestly Vision of Genesis 1. Minneapolis, MN: Augsburg Fortress.
* Wolde, Ellen J. van. 2009. Why the Verb ברא Does Not Mean ‘to Create’ in Genesis 1.1-2.4a. Journal for the Study of the Old Testament 34 (1):3-23.