The word “covenant” in the Old Testament has a precise and consistent meaning. It doesn’t just mean ‘relationship’, as is commonly assumed today. Rather, “covenant” refers to a very specific type of relationship. The best definition of the word, which accounts for all of its uses in the Bible, is as follows:
Covenant = an elected, as opposed to natural, relationship of obligation under oath.
Here is a brief summary of some scholarly work on the definition of the word.
(This post is part of a series. See here for an introduction to the series.)
For more than a century, there has been a scholarly debate over the precise meaning of the Hebrew word בְּרִית (“covenant”). We can summarise the discussion according to four axes: “obligation”, “solemnity”, “relationship” and (more recently) “election”. Each of these axes has been emphasised to a greater or lesser extent in the scholarship under review here.
Earlier understandings tended to play the first three axes off against each other. Wellhausen, Kittel and Gressman viewed a covenant as essentially a set of obligations (usually reciprocal). For Kraetzschmar (1896), it was a means of solemnising such obligations. Pedersen (1914) saw the covenant more as a relationship, with ensuing obligations. This view became somewhat influential, but was seriously challenged by Kutsch’s Verheißung und Gesetz (1973). Kutsch argued that בְּרִית always means “obligation” or “duty”, and does not denote the making of a relationship as such. While בְּרִית can mean “treaty” when used of covenants among humans (i.e. a bilateral acceptance of obligations), when used of Yahweh, בְּרִית never means “agreement” but only unilateral obligation, either self-imposed by Yahweh (a covenant of grace), or imposed by Yahweh on the people (a covenant of law). Kutsch’s view influenced Weinfeld, although Weinfeld also highlights the significance of the common hendiadys between “covenant” and “oath” (i.e. solemnisation).
Barr sought to bring some order to the discussion by calling for the application of functional semantics. For Barr, בְּרִית can be used for a wide range of concepts (expressed in English by various words such as “agreement”, “treaty”, “contract”, “promise”, “obligation”), with solemnity perhaps being the common factor. However, בְּרִית is remarkably restricted in the contexts in which it may occur. A בְּרִית may be made, kept, broken, left, remembered or forgotten; it has certain material or visible signs such as the ark, the book, the tablets and the blood; and it is usually “forever”. That is about all the Biblical writers do with the word. Nobody ever counts or numbers covenants. Nobody ever loves, meditates upon or rejoices in a covenant (except in the later Qumran literature). Nobody ever recounts or retells a covenant, as they do God’s other mighty acts and words. In the light of the New Perspective, we might add here that nobody in the Bible ever “gets in” or “stays in” a covenant either.
Dumbrell, returning to an older view but modifying it somewhat, sees the relational factor in a covenant as foundational and pervasive. He argues that a covenant does not initiate a relationship. Rather “[w]hat the covenant does is formalize and give concrete expression to a set of existing arrangements [. . . T]he covenant will give firm quasi-legal backing to an arrangement which is already in existence”. In other words, a covenant ceremony adds obligation and solemnity to a pre-existing “covenantal” relationship. He cites the various “human” covenants (Gen 21:22–32, 26:26–33; Josh 9; 1 Sam 11:1–3, 18:3; 2 Sam 3:12–21, 5:1–3; 1 Kgs 20:31–34; 2 Kgs 11:17) as examples. On this basis he posits a “covenant with creation” that exists prior to its formalisation with Noah. This would seem to provide a licence for Dumbrell to conceive the whole of reality in covenantal terms.
The problem with Dumbrell’s view is that it replaces the specific biblical usage of the word “covenant” with a more wide-ranging meaning that could, in theory, cover any kind of relationship. It cannot, of course, be disputed that when a covenant is made between two parties, there is some prior “relationship” between these two parties (how or why would two complete strangers spontaneously make a covenant?). However, covenants achieve much more than merely “legalising” existing sets of relationships. For example, the covenant between Joshua and the Gibeonites (Josh 9:15–20) does not merely formalise a pre-existing relationship, it creates a new relationship of peace where previously there had been enmity, even if that enmity was hidden when the covenant was made. Hence “covenant” is not simply synonymous with “relationship”; a covenant is a specific type of relationship.
Hugenberger provides a comprehensive and integrated inductive definition of בְּרִית in his defence of the “covenantal” nature of Old Testament marriage. Decrying “the now discredited notion that ‘covenant [בְּרִית]’ is essentially a synonym for ‘relationship’”, Hugenberger maintains that there is now a “substantial scholarly consensus” about “the major elements which typically comprise a covenant”. His working definition of a בְּרִית in its normal sense is “an elected, as opposed to natural, relationship of obligation under oath”. This definition of “covenant” is not apriori, but comes from a careful examination of the occurrences of the word in their ot context. We have good reason, therefore, to adopt it as our working definition.
A covenant is elected because it is always entered into by choice rather than necessity. Hence “בְּרִית is nowhere employed of naturally occurring relationships and the ordinary obligations which attend them, such as those which exist between parents and a child”. A covenant is relational because it always involves two parties. This explains the common use of familial terminology to describe a covenant, even though the parties previously had no kinship bond. But a covenant is not just any type of relationship; it is specifically a relationship of obligation, because it always binds one or both of the parties to certain specified duties. Finally, a covenant is always solemnified by an oath or oath-sign (e.g. a solemn rite). In fact, בְּרִית may be used as a shorthand for אוֹת בְּרִית, “covenantal sign” (e.g. circumcision, Gen 17:9–14).
 Based on Gordon P. Hugenberger, Marriage as a Covenant: A Study of Biblical Law & Ethics Governing Marriage, Developed from the Perspective of Malachi (Vetus Testamentum Supplements; Leiden: E. J. Brill, 1994), 11.
 Ernest W. Nicholson, God and His People: Covenant and Theology in the Old Testament (Oxford: Clarendon Press, 1986), 13.
 Nicholson, God and his People, 16.
 Nicholson, God and his People, 18–20.
 Nicholson, God and his People, 90.
 Nicholson, God and his People, 91–92.
 Nicholson, God and his People, 106–8.
 M. Weinfeld, בְּרִית”, TDOT 2:253–79 (esp. 255). See also H. Hegermann, “διαθήκη”, EDNT 1:299.
 Weinfeld, TDOT 2:256.
 James Barr, “Some Semantic Notes on the Covenant”, in Beiträge zur Alttestamentlichen Theologie: Festschrift für Walther Zimmerli zum 70. Geburtstag (Göttingen: Vandenhoeck & Ruprecht, 1977), 23–38 (31).
 Barr, “Semantic Notes”, 33.
 Barr, “Semantic Notes”, 34.
 This language comes from Sanders, Paul and Palestinian Judaism, 17.
 William J. Dumbrell, Covenant and Creation: An Old Testament Covenantal Theology (Exeter: Paternoster, 1984), 16–20.
 Dumbrell, Covenant and Creation, 22–26. Dumbrell also employs a semantic argument, distinguishing the perpetuation (הֵקִים) of a covenant relationship from its establishment (כָּרַת). The appropriateness of this distinction had earlier been questioned by Barr, “Semantic Notes”, 33. Paul R. Williamson, “Covenant”, in Dictionary of the Old Testament: Pentateuch (ed. T. Desmond Alexander, David W. Baker; Downers Grove: IVP, 2003), 139–55 (142–43) demonstrates the semantic weakness of Dumbrell’s case.
 Hugenberger, Marriage, 4.
 Hugenberger, Marriage, 11, emphasis mine.
 Hugenberger, Marriage, 180; see also Nicholson, God and his People, 215–16.
 Hugenberger, Marriage, 176.
 Hugenberger, Marriage, 177–79.
 Hugenberger, Marriage, 181. According to Hugenberger, there is some debate about whether a covenant can be entirely “unilateral”, such that only one of the two parties is under obligation; but Hugenberger’s own case is not affected by this debate because he is interested in marriage, which is always bilateral.
 Hugenberger, Marriage, 182–83.
 Hugenberger, Marriage, 173–74.