HistoricalChristian.Faith

Matthew’s formula quotations: do they evidence direct use of a Hebrew (non-Septuagintal) text?

Type: supporting argument (off-timeline) · Assessment: sound — but the strong form (“Matthew translated fresh from a Hebrew scroll”) overreaches the evidence. Matthew’s distinctive “formula quotations” — the fulfilment citations introduced by “that it might be fulfilled which was spoken by the prophet” — are genuinely non-Septuagintal where the Greek and the Hebrew part ways, and at Matthew 2:15 the citation follows the Hebrew of Hosea 11:1 (“my son,” singular) against the Septuagint’s “his children.” That is the standing counter-case that keeps this doctrine’s verdict honest: the New Testament is not uniformly the Septuagint, and Second-Temple Judaism’s text was plural. What holds firmly is the negative claim (these citations are not the Old Greek); what is contested is the mechanism — a fresh translation from the Hebrew, a Septuagint already revised toward the Hebrew, a testimony-collection, or Matthew’s own working. And the three citations usually cited together do not all belong to one class: Hosea 11:1 is MT-distinctive (Hebrew against Greek), Jeremiah 31:15 at Matthew 2:18 is neither (a free form owing to no surviving text exactly), and the “Nazarene” of Matthew 2:23 is a Hebrew word-play, not a text-form citation at all.

Related arguments: the Hebrew (Masoretic) text-form — see also Isaiah 53:4 at Matthew 8:17 (a Matthew formula quotation) · Zechariah 12:10 at John 19:37.

See also: the contrasting species that dominates the doctrine — the New Testament following a distinctively Septuagintal reading of a stable Hebrew — is the Septuagint’s interpretive gloss group (Isaiah 7:14 · Psalm 40:6 · Amos 9:11–12); the third species — the Greek preserving an older Hebrew the Masoretes lost — is Deuteronomy 32:43 at Hebrews 1:6 (with its christologically-neutral twin at Goliath’s height). This row is the fourth and last species: the New Testament standing closer to the Hebrew than to the Greek. The big-picture Septuagint-dependence thesis it counterbalances is weighed on “did the New Testament authors quote the Septuagint?” (assessment sound). This page adds what those do not carry: the adjudication of Matthew’s fulfilment citations as the doctrine’s diversity counterweight.

The claim

“In this it is to be noted that wherever the Evangelist, whether on his own account or in the person of our Lord the Saviour, quotes the testimony of the Old Testament, he does not follow the authority of the translators of the Septuagint but the Hebrew. Wherefore these two forms exist, ‘Out of Egypt have I called my son,’ and ‘for he shall be called a Nazarene.’”

— Jerome, Lives of Illustrious Men 3 (c. 393), on Matthew’s gospel.¹ Jerome is the ancient champion of the reading this argument serves: that Matthew’s fulfilment citations follow the Hebrew rather than the Greek, and so prove the New Testament is not uniformly Septuagintal. His example texts are exactly the two the modern discussion returns to — “Out of Egypt I called my son” (Hosea 11:1 at Matthew 2:15) and “he shall be called a Nazarene” (Matthew 2:23). The interpretation this page weighs is the sober core of Jerome’s thesis: that Matthew’s distinctive fulfilment quotations are the standing counter-case to any “the New Testament = the Septuagint” picture — textual plurality, not Greek-over-Hebrew uniformity. (Jerome’s universal form of the claim — “wherever” Matthew quotes he follows the Hebrew — is itself an overstatement, addressed below and flagged on his detail page.)

Who advances it

The observation is old and the modern form is near-consensus. Jerome (above), defending his new translation “from the Hebrew,” pressed Matthew’s Hosea citation as proof that the apostles used the Hebrew: “Let those who deny the authenticity of the Hebrew copies show us this passage in the Septuagint, and when they have failed to find it, we will show it them in the Hebrew.”² The modern text-critical study of the point begins with Krister Stendahl’s The School of St. Matthew and Its Use of the Old Testament (1954/1968), which isolated the datum every later treatment builds on: Matthew’s formula quotations (the Reflexionszitate, the editorially-inserted fulfilment citations) have a peculiar, non-Septuagintal text-form, distinct from Matthew’s other quotations — the ones shared with Mark, embedded in dialogue and discourse — which are largely Septuagintal.³ The standard surveys carry the same classification: Gleason Archer & Gregory Chirichigno list Matthew 2:15 among the quotations that follow the Hebrew against the Greek, and R. T. France’s commentary treats the formula quotations as the deliberate, Hebrew-attentive core of Matthew’s scriptural method. The one live internal dispute is over mechanism: M. J. J. Menken, in Matthew’s Bible: The Old Testament Text of the Evangelist (2004), argues the formula citations reflect not a fresh translation from the Hebrew but an already-revised Septuagint — a Greek text corrected toward the Hebrew and improved stylistically — which the evangelist quoted from a continuous text.⁴

The evidence for

Hosea 11:1 at Matthew 2:15: the clean non-Septuagintal case. The Hebrew of Hosea 11:1 reads מִמִּצְרַיִם קָרָאתִי לִבְנִי (mimmiṣrayim qārāʼtî li-bnî), “out of Egypt I called my son” — singular, first-person. The Septuagint reads it collectively: ἐξ Αἰγύπτου μετεκάλεσα τὰ τέκνα αὐτοῦ, “out of Egypt I called his children” (the translator having read the consonants as a plural with a third-person suffix). Matthew 2:15 reads ἐξ Αἰγύπτου ἐκάλεσα τὸν υἱόν μου, “out of Egypt I called my son” — which is the Hebrew, not the Greek. Whatever else is true, Matthew cannot have taken this citation from the Old Greek: he has the singular “my son” that carries the whole christological point (Jesus as God’s son recapitulating Israel’s exodus), and the Septuagint does not.²

The formula quotations are text-critically a class apart. Stendahl’s enduring result is not about one verse but about a pattern: sort Matthew’s Old-Testament citations into those he inherited from his sources and those he inserted himself with a fulfilment formula, and the two groups behave differently against the Greek. The inherited quotations track the Septuagint; the inserted fulfilment quotations do not — they show a mixed, Hebraizing form that repeatedly stands with the Hebrew where the Greek diverges. That internal contrast is hard to explain if Matthew simply worked from a Greek Bible; it is exactly what one expects if his distinctive proof-texts were shaped with an eye on the Hebrew (or a Hebraized Greek). This is why Matthew, uniquely among the evangelists, is the doctrine’s diversity counterweight: the same gospel that quotes the Septuagint in its narrative reaches past it in its proof-texts.³

The ancient witness read it exactly this way. Jerome did not merely assert the point; he built a running argument on it. In his Commentary on Matthew and again on Hosea he challenges his critics to find “Out of Egypt I called my son” in the Septuagint, notes that they cannot, and concludes that “there is no doubt that Matthew took this testimony according to the Hebrew truth” — even fending off the emperor Julian, who had mocked the citation as a Christian sleight-of-hand. On Matthew 2:18 (Jeremiah 31:15) he is sharper still: the verse is quoted “neither according to the Hebrew nor according to the Seventy… the Evangelists and Apostles did not follow the interpretation of any translator, but rather expressed in their own words, as if Hebrews, what they read in Hebrew.” That is a first-hand, fourth-century reader of both text-forms testifying that Matthew’s fulfilment citations are not the Greek.⁵

Adversarial assessment

What holds. The load-bearing negative claim is secure and is near-consensus: Matthew’s formula quotations are not the Old Greek Septuagint, and Matthew 2:15 in particular follows the Hebrew of Hosea 11:1 against it. So the catalog’s “MT-distinctive” tag on that row stands, and the broader point the row exists to make is correct: the New Testament is not uniformly Septuagintal, and any tally that reads “the apostles quoted the Greek” as “the apostles quoted only the Greek” is refuted by Matthew’s own practice. This is the honest diversity counterweight the doctrine needs — the mirror image of the gloss cases, where the New Testament turns on a distinctively Greek reading.

What must be refined, so the claim is not over-read.

1. “Non-Septuagintal” is proven; “translated fresh from the Hebrew” is not. The data fix that Matthew’s formula citations diverge from the Old Greek toward the Hebrew; they do not fix how that happened. At least four mechanisms fit: (a) Matthew (or a source) rendered the Hebrew directly; (b) he used a Greek text already revised toward the Hebrew — the very phenomenon the kaige / proto-Theodotion recensions represent, and Menken’s reconstruction of Matthew’s Bible; (c) he drew on a pre-existing collection of Christian testimonia (a mechanism weighed on its own at Were the composite citations drawn from a pre-formed testimonia collection?); (d) he adjusted a Greek text ad hoc to sharpen the fulfilment. At Hosea 11:1 the point is pressed by the later Jewish revisers — Aquila, Symmachus, and Theodotion also read a singular “my son” there — so Matthew’s agreement with the Hebrew is also agreement with a Hebraizing Greek tradition, and cannot by itself decide direct Hebrew use. The strong, Jerome-style form (“the apostle translated the Hebrew”) is therefore a possible reading, not a demonstrated one. What the tag reports — and all it should — is the text-form (non-Septuagintal, Hebrew-aligned), not the translator’s workshop.⁴

2. Jerome’s “wherever” is too much; the phenomenon is confined to the fulfilment citations. The claim that Matthew always follows the Hebrew is false as stated: his non-formula quotations — the ones he shares with Mark, in controversy and discourse (e.g. Matthew 4:4’s “man shall not live by bread alone,” or the great-commandment texts) — are largely Septuagintal. The honest statement is Stendahl’s, not Jerome’s: it is the formula quotations that are the class apart. That confinement does not weaken the argument; it sharpens it, because the contrast within one gospel between Septuagintal narrative-quotations and non-Septuagintal proof-texts is itself the strongest evidence that the proof-texts were handled with special, Hebrew-attentive care. Jerome’s overreach is flagged on his detail page.¹

3. The three citations are three different things; only Hosea 11:1 is “MT-distinctive.” Jerome’s own pairing bundles unlike cases. Hosea 11:1 (Matthew 2:15) is a real MT-against-LXX divergence — the clean case. Jeremiah 31:15 at Matthew 2:18 is different: Matthew’s Rachel-lament matches neither the Septuagint nor the Masoretic Text verbatim — it is a free, abbreviated rendering (Jerome: “neither according to the Hebrew nor the Seventy”), Hebraizing in feel but not a copy of any surviving text, so it belongs in the “neither / free” class, not the MT-distinctive one. And Matthew 2:23’s “he shall be called a Nazarene” is not a text-form citation at all: it matches no identifiable Old-Testament sentence (Matthew signals as much by writing “the prophets,” plural). Its best anchor is a Hebrew word-play — on נֵצֶר (nēṣer, “branch/shoot”) of Isaiah 11:1, the messianic sprout “out of the stem of Jesse,” or on נָזִיר (nāzîr, the consecrated one). A pun that works only in Hebrew evidences the evangelist’s Hebrew competence — the Septuagint renders nēṣer by ἄνθος (“flower”) and loses the sound-link entirely — but it is not a quotation whose “text-form” can be Septuagintal or Masoretic, so its catalog tag stays ambiguous. So of the three, one lands MT-distinctive, one neither/free, one a Hebrew pun: the bundle proves Hebrew attention across Matthew’s fulfilment citations, not a single uniform text-form.⁵

4. This is a counter-weight, not a counter-verdict. Matthew’s formula quotations keep the doctrine’s verdict two-part; they do not overturn it. The wider New-Testament pattern — the divergence proof-texts in Acts, Hebrews, Romans, 1 Peter, weighed across this catalog — still leans Septuagintal where Greek and Hebrew part ways, and the honest headline stays “a plural text, quoted plurally,” not “the New Testament followed the Hebrew.” Matthew is the one evangelist who supplies the systematic exception, which is precisely why the doctrine needs him on the page: he is the evidence that the Septuagintal lean is a lean, not a law.³

Assessment

Sound. The interpretation this argument serves — that Matthew’s formula quotations are non-Septuagintal where the Greek and Hebrew diverge, following the Hebrew at Hosea 11:1, and so stand as the doctrine’s diversity counterweight against any “the New Testament = the Septuagint” picture — holds, and is near-consensus from Stendahl onward. The catalog’s “MT-distinctive” tag on Matthew 2:15 is correct: the singular “my son” is the Hebrew’s and not the Greek’s. Three refinements keep the claim from overreaching: (a) what is proven is the text-form (non-Septuagintal, Hebrew-aligned), not the mechanism — a fresh Hebrew translation, a Hebraizing/revised Greek (Menken), or a testimonia source all fit, and Matthew’s agreement with the Hebrew is also agreement with Aquila/Symmachus/Theodotion, so Jerome’s “the apostle translated the Hebrew” is possible, not demonstrated; (b) the phenomenon is confined to the fulfilment citations — Matthew’s other quotations are largely Septuagintal, so Jerome’s universal “wherever” is an overstatement (flagged on his detail page), and the real force of the argument is the contrast within one gospel; and (c) the three verses usually cited together fall into three classes — Hosea 11:1 MT-distinctive, Jeremiah 31:15 neither/free, the “Nazarene” a Hebrew-only word-play — so the bundle evidences Hebrew attention, not one uniform Hebrew text-form. Kept to that, Matthew is the honest counter-case that makes the doctrine’s verdict truthful: not “Greek over Hebrew” but “a plural text, quoted plurally,” with the one evangelist who reaches past the Septuagint in his proof-texts standing as the proof that the Septuagintal lean elsewhere is a preference and not a rule.

Sources