How the New Testament Reads the Old

Everything this course has taught you so far pushes in one direction. A text means what its author was doing with his words in his own setting. A reading that the writer could not have recognised is a reading under suspicion. You have been trained to ask what Hosea was saying to eighth-century Israel before you ask what Hosea says to you.

Then you open Matthew 2. Hosea writes, 'When Israel was a child, I loved him, and out of Egypt I called My son.' It is a past-tense sentence about the exodus, and the next verse says that this son promptly went off and sacrificed to the Baals. Matthew takes the line, applies it to a toddler coming back from Egypt roughly seven centuries later, and says this happened to fulfil what the Lord spoke by the prophet. On the method you have been taught, that looks less like exegesis and more like the thing you have been warned against.

There are two easy exits and both are closed. The first is to insist that Hosea was secretly predicting the Messiah all along, so that Matthew is simply reporting a prophecy come true. You cannot get there from Hosea 11 without doing violence to the chapter, and the dishonesty will be obvious to any student who reads the chapter. The second exit is to concede that Matthew was an enthusiastic first-century preacher making free with his sources, and that his citations are decoration rather than argument. That exit costs more than it looks. Peter's Pentecost sermon turns on Psalm 16 read as speaking of someone other than David. Hebrews builds a priesthood out of Psalm 110 and eight words about Melchizedek in Genesis 14. Paul argues for the right of ministers to be paid from a law about oxen. Pull the apostles' handling of the Old Testament out of the New Testament and a great deal of Christian doctrine comes out attached to it.

So the question is live, and it is the hinge on which this whole module turns. Either the apostles were doing something legitimate that you can learn, or they were doing something legitimate that you cannot copy, or they were doing something illegitimate whose conclusions you have inherited anyway. Serious Christian scholars hold all three, and they are still arguing. What follows works three hard cases first, because a position taken before the evidence is felt is worth nothing.

Case one, Hosea 11:1 in Matthew 2:15. Start with a detail that is easy to miss and that cuts against the lazy explanation. The Septuagint renders the Hebrew singular 'My son' as a plural, 'his children', which makes the verse unmistakably about the nation. Matthew does not follow the Septuagint here. He follows the Hebrew singular. Whatever he is doing, he is not accidentally inheriting a Greek mistranslation; he has chosen the singular, and the singular is the whole point of his use.

Now read Hosea 11 to the end rather than stopping at verse 1, because the chapter does not stay in the past. Verse 11 says of a scattered people that 'they will come trembling like birds from Egypt and like doves from the land of Assyria. Then I will settle them in their homes, declares the LORD.' Hosea has taken the exodus, the event of verse 1, and used it as the shape of a rescue still to come. This is Greg Beale's central argument, laid out most fully in an article in the Journal of the Evangelical Theological Society 55/4 (2012), pages 697 to 715, which is free to read on the Evangelical Theological Society's own site. His claim in that article is that Matthew's reading was stimulated by Hosea's own typological understanding of the verse. Matthew is not lifting a stray line out of a chapter that has no future in it. He is reading a chapter that has already turned its own history into a pattern, and he is identifying the one in whom the pattern comes to its point. Add to this that Israel is called God's firstborn son in Exodus 4:22, in the exodus narrative itself, and the sonship language is not Matthew's invention either.

That is the strongest case for Matthew. Notice what it does not do. It does not show that Hosea, writing in the eighth century, was referring to Jesus of Nazareth. It shows that Matthew is working with a real feature of the text rather than inventing a hook. Those are different claims, and a teacher who slides from the second to the first has stopped being honest.

Case two, Jeremiah 31:15 in Matthew 2:18, is harder, and you should resist anyone who makes it easy. Jeremiah's line is Rachel weeping for her children at Ramah, refusing to be comforted, because they are no more. Ramah was a staging post for deportation; Jeremiah 40:1 puts the prophet himself there in chains among the exiles. The verse is the low point of a passage that immediately reverses itself: verses 16 and 17 tell Rachel to stop weeping, because the reward for her work will come and her children will return from the land of the enemy. Matthew quotes the weeping and applies it to children who do not come back.

The best defence is not a trick. It is the observation that ancient writers commonly cite a line to invoke the passage it comes from, and that the passage here is Jeremiah 31, the new covenant chapter, ending in verses 31 to 34 with the promise that God will put his law in their minds and inscribe it on their hearts. On this reading Matthew places the massacre at Bethlehem inside a chapter that already contains both the grief and the consolation, and the consolation arrives in the one child who got out. C. H. Dodd argued along these lines in 1952, in a short book called According to the Scriptures, which is still the starting point for the whole discussion. The objection is equally clear: this asks a great deal of Matthew's readers, and the surface of his sentence uses the grief alone. Whether the wider context is genuinely in play or is a modern rescue attempt is exactly what is disputed.

Case three, Galatians 4:21-31, is the one where the apostle himself tells you what he is doing, and the word he uses is the problem. Paul takes Abraham's two sons, one by the slave Hagar and one by the free Sarah, and says in verse 24 that these things are allegoroumena, a Greek participle from the verb that gave us the word allegory. The BSB renders the sentence 'These things serve as illustrations', which is softer than the Greek verb and so quietly leans toward the typological side of the dispute below. Translations that bring the word allegory into English (the ESV is one; look it up on Bible Hub) lean the other way, so check the Greek before you trust either. He then maps the slave woman onto Mount Sinai and the present Jerusalem, and the free woman onto the Jerusalem above and the churches he is writing to. He tells Gentile Christians who are being pressed towards the law that they are Sarah's children, and he puts the law-keeping Jerusalem of his own day on Hagar's side of the line.

Sit with how startling that is. Any first-century Jewish reader would have mapped the story the other way round, with Israel as Isaac and the Gentiles as Ishmael. Paul reverses it. Nothing in Genesis identifies Hagar with Sinai, and Paul's remark in verse 25, which the BSB gives as 'Now Hagar stands for Mount Sinai in Arabia', is geographically odd and textually unstable, with the manuscripts differing over the wording. The BSB's 'stands for' is itself a smoothing, since the Greek verb is simply 'is'. There is no obvious historical correspondence being traced here, of the kind that a defender of typology can point to. Paul appears to be reading a second sense out of the words, and he names his own procedure with the verb that gave us the word allegory. Longenecker classes this as allegorical exegesis of the kind practised in the synagogue of the day and says plainly that we should not imitate it. C. K. Barrett suggested that Paul is working with a text his opponents had brought into the argument and is turning it against them, which would make the move polemical and occasional rather than a general method. Others argue that the Greek verb at this date covers figurative reading broadly, including what we would call typology, and that Paul's argument does rest on real persons and a real history. The lexicon will not settle it for you. That is not a gap in your education; it is the state of the question.

Richard Longenecker set the terms of the modern debate with Biblical Exegesis in the Apostolic Period, published in 1975 and issued in a second edition in 1999 whose new preface asks directly whether we can reproduce the exegesis of the New Testament. His account, building on Krister Stendahl's 1954 study of Matthew, is that the New Testament writers used the recognised exegetical techniques of their own world: literalist, midrashic, pesher, and allegorical readings of the sort attested at Qumran and in the rabbis. His conclusion is the sentence everyone quotes. Christians today are committed to the apostolic faith and doctrine of the New Testament, he argues, but not necessarily to the specific apostolic exegetical practices. Note the word specific, which is in his sentence and is often dropped when he is paraphrased; he is not saying that no apostolic procedure can be followed, but that the particular techniques of pesher and midrash cannot. We cannot reproduce the revelatory stance of pesher interpretation or the atomistic handling of midrash, and he says we should not try. Read charitably, this is not a low view of scripture. It is a high view of apostolic authority combined with modesty about our own.

Greg Beale answered him directly in an article whose title asks whether Jesus and his followers preached the right doctrine from the wrong texts, in Themelios 14.3 (1989), free to read in the Themelios archive. Beale's case runs in three moves. First, the apostles' readings, examined closely, respect the contexts of the passages they use far more often than the critics allow, and the alleged violations tend to dissolve on closer reading, as with Hosea 11. Second, the apostles argue to be understood; Paul expects the Galatians to follow him, which means his method must have been publicly followable rather than privately revealed. Third, and this is the load-bearing one, if their conclusions are valid and their method invalid, then conclusions are being received from a procedure that cannot bear them, and the authority of the conclusions becomes bare assertion. His own formulation on the disputed point is that it is not necessary to claim that we must have the apostles' inspiration in order to reproduce their method.

Beale has his critics too, and the section would be dishonest without them. Longenecker returned to the question in the preface to his second edition and was not persuaded, holding that the readings Beale rehabilitates have been rehabilitated by being described in categories the apostles themselves do not use. Peter Enns presses the reverse objection: that readings of this kind smooth away the genuinely Second Temple character of apostolic exegesis in order to make it acceptable, and that the historical evidence for how Jews of the period actually read their scriptures is the first casualty. And Walter Kaiser's objection applies to Beale as much as to anyone: showing that Hosea 11 contains a forward-looking trajectory still does not show that Hosea 11:1 is a prediction, so the word fulfil is being asked to carry a sense it does not carry in ordinary prophetic usage.

A related position, and the one that supplies the vocabulary most preachers now use, is that Jesus recapitulates the history of Israel. The formulation belongs to R. T. France, in Jesus and the Old Testament, published in 1971, and it is a typological rather than a predictive account: Matthew's citation formula marks a correspondence between Israel's story and the story of the one who embodies Israel, and it does not claim that Hosea forecast the event. Applied to Matthew 2:15 the reading is natural, because Israel is called God's firstborn son in the exodus narrative itself. The objection is the one Kaiser makes above in a different form. Recapitulation is a theological judgement about what Israel's history was for, and a judgement of that kind cannot be read off Hosea 11 by inspection, so the reading relocates the problem into the doctrine of history rather than dissolving it.

The sensus plenior proposal takes a different tack and grants the difficulty. The Latin phrase means fuller sense. The Spanish Jesuit Andres Fernandez, sometimes miscopied as Andrea in English-language footnotes, coined it in 1927; Raymond Brown gave it its standard form in a 1955 dissertation and in articles through the fifties. Brown's definition is careful: an additional, deeper meaning intended by God but not clearly intended by the human author, which is seen to exist in the words of a text when they are read in the light of further revelation. Stated at its strongest, the proposal is not a licence but a restraint. It concedes that Hosea meant Israel, which is what the grammar and the history show, and it refuses to pretend otherwise; it locates the surplus in the intention of the divine author rather than in the ingenuity of the reader; it makes the apostles' readings disclosures rather than techniques, so that nobody may run the procedure at home; and it explains, as no other proposal quite does, why the New Testament's readings feel both surprising and right. Its critics are real. Walter Kaiser argues, most fully in The Uses of the Old Testament in the New in 1985, that a fuller sense not available to the author is not a meaning of the text but an addition to it. Douglas Moo, in the 1986 volume Hermeneutics, Authority, and Canon, argues that what the proposal reaches for is better described as a canonical relation between texts. And Brown himself grew more cautious about the category's usefulness in later work.

A further position is typology, treated in the next lesson at length. Leonhard Goppelt's study Typos, published in German in 1939 and in English in 1982, argued that the New Testament's characteristic way of reading the Old is neither prediction nor allegory but the tracing of historical correspondence, with escalation, between earlier persons, events and institutions and their fulfilment in Christ. If typology rather than allegory is the primary operation, then most of the hard cases are cases of a recognisable and teachable procedure. The standing objection, which goes back to Herbert Marsh and which the next lesson works through, is that Goppelt's criteria describe correspondences without telling you which correspondences God intended, and that is the only question that matters.

A last position, currently the most influential in academic work, says that the apostles read whole storylines rather than isolated verses, and that the argument has been distorted by treating citations as proof texts. Dodd began it. Richard Hays developed it with the idea of metalepsis, where a quoted fragment brings its unstated context with it, in Echoes of Scripture in the Letters of Paul in 1989 and Echoes of Scripture in the Gospels in 2016, arguing for reading the Old Testament figurally and retrospectively. Peter Enns, in Inspiration and Incarnation in 2005, argued that what governs the apostles' readings is a christotelic conviction, that the story of Israel has Christ as its end, so that the readings are legitimate as conclusions drawn from that end rather than as procedures run forwards. The objection is sharp and comes from Christopher Stanley, who argues in Arguing with Scripture in 2004 that Paul's largely Gentile congregations cannot have possessed the scriptural competence these readings require, so that the unstated contexts Hays hears are more likely the modern scholar's than the first-century hearer's. Behind that lies the broader worry that an echo nobody can fail to hear once it is pointed out is not thereby an echo the author made, and that the method has no way of being shown wrong. If you want the main options set side by side by their own defenders, the 2008 volume Three Views on the New Testament Use of the Old Testament, edited by Kenneth Berding and Jonathan Lunde, has Kaiser arguing single meaning and unified referents, Darrell Bock arguing single meaning with multiple contexts and referents, and Enns arguing fuller meaning with a single goal.

Some things are not in dispute among serious readers, and you can teach them with confidence. The apostles read the Old Testament backwards from the resurrection; their reading is retrospective, and they say so. Luke 24 has the risen Christ himself teaching that the Law, the Prophets and the Psalms speak of him, which means the New Testament presents its own reading as authorised rather than obvious. The techniques they used were not invented by them: the commentary on Habakkuk found at Qumran, usually cited as 1QpHab, applies prophetic texts to the community's own present in ways formally comparable to New Testament pesher. Their citations vary in form, sometimes following the Hebrew, sometimes the Greek, sometimes neither exactly, which by itself rules out any theory that requires them to be quoting mechanically. And no major Christian doctrine rests on a single odd citation; where the apostles argue for something load-bearing, they argue for it more than one way.

What is not settled is the question you will be asked. Whether a Christian today may read the Old Testament as the apostles read it, or only may believe what they concluded, remains genuinely open, and the people on each side are not fools or liberals or fundamentalists but scholars who have read the same evidence and weighed it differently. Whether a fuller sense intended by God but not by the prophet is a coherent idea is unsettled. Whether Galatians 4 is allegory or typology is unsettled, and probably cannot be settled from the word alone. Whether Matthew 2:18 shows contextual sensitivity or the lack of it is unsettled. You should be able to state each position in a form its defenders would accept, and you should not pretend to a resolution that the field does not have.

There is more agreement about practice than about theory, and that is useful. Nearly everyone in this argument accepts three working rules. Do not build a doctrine on a reading nobody else can check; if your interpretation depends on a connection only you can see, the problem is more likely in you than in the text. Distinguish showing that an apostolic reading is coherent from claiming you could have produced it in advance; the first is exegesis, the second is a claim about yourself. And keep the plain sense doing the work, so that the christological reading you offer is an enrichment of what the passage says rather than a substitute for it.

One last thing to notice, because it changes the tone of the whole discussion. The apostles' readings are not arbitrary in their effect. They consistently find the same thing: that the God who brought a son out of Egypt, who wept over exiles and promised to bring them home, who made a covenant with Abraham, has done that kind of thing again and finally. Whatever you conclude about their method, their conclusion is not a random reading of the Old Testament. It is a reading of the Old Testament's own direction of travel. That observation does not settle the argument. It does tell you why the argument has never been resolved by simply declaring the apostles wrong.