Because I aim to win friends and influence people, I have just published a completely uncontroversial article on Matthew 1:25 and the issue of Mary’s perpetual virginity. It’s freely available here. This paper is basically an accident: I was studying a change in LXX Genesis 8:7 where the translator (against the Masoretic Text) inserted the word until, changed the verb, and then placed it under negation (not…until) to produce an actualization inference that the raven returned to Noah. It occurred to me: these changes are almost identical to the grammar of Matthew 1:25. And so, the paper was born.
If you search the history of interpretation of Matt. 1:25, scholars often discuss the word until but seem completely unaware of the literature in linguistics about it. The better arguments usually go like this: “The word ‘until’ can imply a change, but not necessarily. Matthew says nothing about whether conjugal relations later took place.”
Is that true? No, unfortunately not. It’s true that until can generate two types of inference. But it is not true that both inferences are mere implications that can be cancelled. Simplifying just slightly, when until is combined with a bounded predicate (one that has a clear endpoint) and placed under negation, it generates an actualization inference. In the paper I call this meaning after but not before (I didn’t leave until you called = I left after but not before you called). This inference cannot be cancelled. The grammar (for reasons I explain) entails it. However, when until is combined with an unbounded predicate, or a bounded predicate in a positive sentence, it usually generates a termination inference (I was lost until I took a left turn = I was lost before but not after I took a left turn). This inference can be cancelled. It is not an entailment of the grammar.
I argue the grammar of Matthew 1:25 represents the former, not the latter, and demonstrate the actualization inference after but not before is obligatory. The Greek euphemism ἐγίνωσκεν αὐτὴν (know her) is bounded because like ‘find’ if it happens at all, the event was completed (e.g. I find my friend in the mall, but I didn’t find her is a contradiction). This is important because not…until merely limits the time at which the event in question was not true (hence ἐγίνωσκεν not ἔγνω). Once that time is over, bounded events like ‘find’ culminate instantly (I didn’t find my friend until I called him = after I called him, I found my friend). If the euphemism ἐγίνωσκεν αὐτὴν in 1:25 is like ‘find’, it indeed happened after Mary finished giving birth. Similarly, this is why a better translation of the euphemism is ‘come to know’ rather than merely ‘know’ — English speakers have trouble ‘hearing’ the entailment because the euphemism is not part of our lexicon and we process ‘know’ by default as a state.
It is crucial not to misunderstand the argument at this stage. Under the conditions I describe in the paper, the actualization inference in Matt. 1:25 is not a mere implication. It is part of the truth conditional content of the utterance. It does not depend on speaker intention. It is not a sense of the preposition. It is not unique to a particular lexicalization of until in Greek (side note: the relativizer οὗ does not make anything ‘emphatic’). And it is not less ‘real’ than meaning that is coded explicitly (just try to do the tricks on John 9:18 that people do with Matthew 1:25!). The actualization inference is an entailment carried by the grammar itself.
And although I know you should typically doubt this sort of thing, the conditions I describe above demonstrate that the history of interpretation of Matthew 1:25 is founded on a red herring (the price we pay for ignoring 50+ years of literature in linguistics). In my research, I read everything I could find. To my knowledge, every single counter example in the literature fails to form a minimal pair with Matthew 1:25: they either lack negation, the predicate is unbounded, or both(cf. 2 Samuel 6:23, Matt. 13:33, 1 Cor 15:24-25, et al.). And herein lies the point. The only way to ‘prove’ Matthew 1:25 doesn’t carry an obligatory actualization inference is to confuse it with disanalagous examples. When an example does form a minimal pair with Matthew 1:25 (e.g. LXX Genesis 8:7), the actualization inference is always obligatory.
In one sense, the paper I published should have been unnecessary. Any scholar might have searched and easily found decades of literature on boundary adverbials in linguistics. While fads abound in biblical scholarship, it is no exaggeration to say that a mountain of philological controversies could be avoided were scholars to look to the scientific study of language. Anyone interested in the side quest can check the Jerusalem or Russian Synodal Bible to see how Roman Catholic and Eastern Orthodox translators alter the linguistic facts to prevent readers from drawing the ‘wrong’ inference. On the Protestant side, consider documents like the Westminster Confession (11.4), which use the same grammar as Matt 1:25 to generate the actualization inference: “they are not justified until the Holy Spirit doth, in due time, actually apply Christ unto them” (imagine claiming they are not justified after, either!). The key point here is that we don’t treat our own grammatical choices with equal uncertainty as biblical texts (which is why Greek church fathers can actually be a problem for interpreting Greek grammar).
But if all this is true, you ask, why was Matthew not more explicit? The reason is actually quite simple: what we select for communication is based on principles like informativeness, relevance, and other conversational rules. The not…until construction is used to inform someone about the last moment before a relevant change took place. Matthew uses not…until, rather than another construction, because only the atypical aspect of the marriage deserves explanation: Mary was a virgin when Jesus was born. Matthew believes that conjugal relations did indeed later take place (the relevant change), but he can leave this to inference the absence of a defeater otherwise because it is not informative or relevant to the issue at hand. This is why later in the gospel, despite cries to the contrary (translation! translation!), siblings appear unannounced.
Responses so far have basically taken five forms:
(1) “You’re not an Ancient Greek speaker and all the Greek fathers disagreed with you. Why should I trust you?” I address this response in the paper. It is widely known that native speakers can be highly unreliable when describing their language because they blur description with prescription: their words are valuable “not as pure sources of data, but as instances of metalinguistic performance.” This is nowhere more the case than in controversial matters of theology.
(2) “Here are some examples that disprove your rule.” Every single one either lacks negation, the predicate is unbounded, or both. Others are simply vague (e.g. “I didn’t touch the car until the police arrived” where the plural collective police denotes an unknown amount of arrival-events, leaving the right boundary of the until time span vague). All of these are disanalagous in some way to Matthew 1:25.
(3) “That’s just your opinion.” To be fair, this one is usually more sophisticated: “Your judgment is merely probable, not demonstrative. It does not produce scientia, and since you do not possess the auctoritas of God speaking, I choose to ignore your work in light of the Church’s motives of credibility.” Well, fine. If no grammatical argument can matter unless it’s already ratified by the authority you have chosen to trust, you are only interested in conclusions, not arguments, and I can’t help you there.
(4) “Even the reformers disagree with you.” So what? My view is that of Samuel Rutherford (“not in Scripture”). In the end, it does not matter whose view it is, but whether it is true.
(5) “Why do you hate the holy virgin?” I do not. As John Owen wrote, “Protestants yield to the blessed Virgin all the honour that the Scripture allows them, or direct them unto, or that the Primitive Church did ascribe unto her.” Now, I disagree that the former allows one to affirm her perpetual virginity. But it is not that serious an issue if you do. According to Basil of Caesarea (who affirmed perpetual virginity), “[Matthew 1:25] could make one suppose that Mary, after having offered in all her purity her own service in giving birth to the Lord, by virtue of the Holy Spirit, did not subsequently refrain from normal conjugal relations. That would not have affected the teaching of our religion at all, because Mary’s virginity was necessary until the service of the Incarnation, and what happened afterward need not be investigated in order to affect the doctrine of the mystery” (PG 31, 1468).
Koine-Greek 