Repository path: workshop/translations/beowulf-ingeld/R04-v1/translation.md · rendered 2026-09-09

Beowulf, ll. 2015–2070a — the Freawaru / Ingeld digression, into modern English

Contamination: high. Beowulf is plausibly the most-translated poem in the English language; complete versions by Morris & Wyatt, Hall, Gummere, Kirtlan, Clark Hall, Tolkien, Raffel, Heaney, Liuzza, Headley and dozens more exist, and many are freely available online. Renderings of this passage almost certainly sit in the lead's training data, and the lead cannot establish that the English below is independent of remembered ones. No published rendering of this passage, in any language, was consulted before this translation and its log were frozen — the R04 source-only rule was followed, and the passage was chosen partly because it is one of the less anthologised stretches of the poem. That protects against priming; it does not protect against memory.

One further, partial priming, declared (see log §1): the session's baton named Morris & Wyatt's archaizing as the study limb's foreignisation contrast, so the translator knew before translating that one comparison text is heavily archaic. That knowledge plausibly influenced the decision to write plain modern prose (log §3a). It did not involve reading any of Morris's text.

The translation

The company was in joy. Never in a long life have I seen, under the vault of heaven, greater rejoicing over the mead among men seated in hall. From time to time the renowned queen, the peace-bond of peoples, went the whole length of the hall-floor and urged on the young men; often she gave a ring-band to a man before she went to her seat. And from time to time, before the seasoned retainers, Hrothgar's daughter carried the ale-cup down the line of the nobles — Freawaru, as I heard the men seated along the hall-floor name her, where she gave studded treasure to the fighting men. She was promised — young, gold-adorned — to the gracious son of Froda.

The friend of the Scyldings, the guardian of the kingdom, has brought this about, and counts it sound policy that by this woman he should lay to rest his portion of deadly feuds and quarrels. Time and again, and nowhere seldom, after a prince has fallen the killing-spear is laid aside only a little while, however worthy the bride.

It may well displease the lord of the Heathobards then, and every thane of that people, when he walks the hall-floor with the lady — a Danish lord's son, attended by the retainers — and on him the heirlooms of men long dead are gleaming, hard and ring-marked, the Heathobards' own treasure, for as long as they were allowed to wield those weapons, until they led their dear companions, and their own lives, astray into the shield-play.

Then, over the beer, one who sees a ring will speak: an old spear-fighter, who remembers all of it, the spear-death of men — his heart is grim. Gloomy-minded, he begins to sound the young warrior's spirit through the thought of his breast, to wake the evil of war, and he speaks out this word:

"Can you tell that sword, my friend — the one your father carried into the fight, under his battle-mask, the last time he bore it, the dear iron — there where the Danes struck him down and held the ground of slaughter, when Withergyld lay dead, after the fall of fighting men, the valiant Scyldings? And now, here, the son of one or another of those killers walks the hall-floor, exulting in the war-gear, boasts of the killing, and carries the treasure that you by rights should hold."

So he urges and reminds, time after time, with bitter words, until the hour comes when the lady's thane, for his father's deeds, sleeps blood-stained after the blade's bite, his life forfeit; and the other gets away from there alive, for he knows that country well. Then on both sides the sworn oaths of the nobles are broken, and after that deadly hatreds well up in Ingeld, and his love for his wife grows cooler after the surges of care. So I do not reckon the Heathobards' loyalty sincere towards the Danes, nor their portion in the lordly bond, nor the friendship firm.

Translator's log

Frozen 2026‑07‑25, before any evaluation of this translation was designed, and before any published rendering of the passage was read (charter §3, A4). Decisions and alternatives; not quality claims. The two sentences that shade into evaluation are marked. The page's front matter carries internal-judgment-only: true for the whole file.

1. The passage, and why this one

The session's unit is a second intralingual cell, to test whether A-yosano-yomogiu's findings are about intralingual translation or about Japanese (TH-20260724-translation-distance-axes, "What this does NOT establish"). Old English → modern English is the hard case, and that is the point of choosing it: it is intralingual by name, but the lexicon mostly did not survive, so inheritability and intralinguality come apart here in a way they could not in the Japanese cell, where the script and most of the content words transferred whole.

Within the poem, ll. 2015–2070a were chosen on four criteria:

1. Realia-dense, across a range of inheritability. The passage runs a hall-feast and a feud in 56 lines: mead, ale, beer, rings, a cup, a sword, a helmet, war-gear, oaths, a marriage alliance, a retinue. Some of those words survive into modern English (bēor, ealu, medu, gold, īren), some survive with the sense bent (eorl, wīf, drēam, glæd, rǣdan), and some are simply gone (flet, duguð, bēah-wriða, lind-plega, here-grīma, māððum). That gradient is the study limb's instrument.
2. Threads to hold or drop. flet occurs four times; sibb frames the digression at both ends; dǣl, wine, hæleð, duguð, sealde, hryre, fæder, byre each occur twice. Half-line 2035b and half-line 2055b are verbatim identical (on flet gǣð / on flet gǣð) at the two moments when the same scene is described first by Beowulf and then by the old warrior. That repetition is a countable.
3. Not anthologised. It is not the opening, not Scyld's funeral, not Grendel's approach, not the dragon, not the last survivor's lament. Verbatim memorisation is likelier on those.
4. Voice. It is a speech inside a speech inside a report — Beowulf quoting an imagined old spearman — which gives the translation somewhere for register to move.

2. What was hard, in order of how much time it took

(a) flet, four times, with the modern reflex gone and the obvious calque poisoned.

flet is the floor of the hall, and by extension the hall as a social space. It appears at 2018 (flet eall geond-hwearf — the queen crosses it), 2023 (flet-sittende — the men on it), 2035 (on flett gǣð — the Danish escort walks it beside the bride) and 2055 (on flet gǣð — the same man walks it wearing the dead Heathobards' gear). The last two are the same half-line. In the source, the repetition is the argument: the old warrior is quoting Beowulf's own picture back at him with the sympathy reversed.

Modern English has no reflex of flet — the word is dead — so the intralingual translator here is in exactly the position of the interlingual one: the option to copy does not exist. Worse, the natural calque is booby-trapped. "The floor" in modern English means the ground you stand on and, in the phrase "on the floor," carries a boxing-ring or a dance-hall or a legislature. "He goes on the floor with the woman" would be actively wrong.

Options: (i) "the hall" at all four sites, which is idiomatic and loses the concreteness — the source says floor, and the floor is what a queen physically walks the length of; (ii) vary — "the hall," "the men in hall," "walks in," "comes in" — which reads best and destroys the two repetitions; (iii) manufacture a compound and hold it.

Chose (iii): "hall-floor" at all four sites, 4/4, with 2035 and 2055 rendered by the identical phrase "walks the hall-floor." The cost is that I had to build a compound the target does not have, to hold a thread the source held with a plain noun; "hall-floor" is a slightly archaeological word in modern English and there is no way to make it invisible. Paid knowingly.

(b) wīf — one Old English word, two modern words, and the split is forced.

At 2029 Hrothgar plans to settle the feud mid þȳ wīfe; at 2066 Ingeld's wīf-lufan cools. Old English wīf covers both: it is woman, and it is woman-as-spouse, and the poem does not distinguish them because the passage's whole subject is a woman turned into an instrument of policy. Modern English has no word that spans the two. "With that wife" is wrong at 2029 (she is not yet married, and the point is that she is a means); "his love for his woman" is wrong at 2066 (it imports a modern register the line does not have).

Rendered as "by this woman" and "his love for his wife." That is a forced explicitation: my English states at both sites which of the two things is meant, and the source states neither, and the movement between them across forty lines is precisely what the source is about. There was no live alternative — this is the loss, not a choice among losses. It is the same shape as the Genji honorific case (T-genji-yomogiu-R04-v1 log §4), reached from a different direction: there a category the target lacked forced me to name agents; here a semantic distinction the target obligatorily makes forced me to name a relationship.

(c) The false friends, refused one at a time.

Modern English holds the descendant form of several of this passage's key words and has bent every one of them. In each case the modern word was available, and in each case using it would have been a misreading:

wine is the sharpest of these and the most nearly comic: the word means friend, protector, beloved lord, it occurs twice, and both occurrences are at a drinking feast, so the modern reader's eye is offered a reading that is not merely wrong but locally plausible. glæd and gladiað are the same root at 2026 and 2037 and mean two different things there (gracious, of a prince; gleam, of metal), neither of which is what "glad" means; the source's own reader heard the two as one word, and I could not.

Note against the Genji cell: this is the うるはし/麗しい mechanism, in a second language, at scale — nine instances in 56 lines rather than one thread in 951 characters. But there is a difference worth logging, and it goes the other way. In the Genji passage the drifted word was a tempting candidate that had to be refused at each site. Here the drift is mostly so violent that no temptation arises: nobody would render drēam "dream". Only wīf→"wife" and eorl→"earl" are close enough to actually tempt. So drift creates false friends only inside a window; too little drift and the word is simply usable, too much and it is obviously unusable.

(d) The realia fork, and how often it was actually forked.

I went through the passage's culture-bound items expecting, from A-yosano-yomogiu, that an intralingual pair would let most of them stand. It did not. The tally, by what modern English made available:

• Free copy, no decision at all (the modern word is the Old English word, still in use, still meaning the same): medu → mead, ealu → ale, bēor → beer, gold → gold, īren → iron, sweord-class words like mēce excepted. Five or six items, all of them drink and metal.
• Copy possible but archaic-marked: þegn → "thane". The word exists in modern English only as a historicism, so copying it colours the sentence in a way the source's plain word does not. I copied it, and the colour is a cost I accepted.
• No copy available — a real fork, exactly as in an interlingual pair: flet (2a), duguð → "the seasoned retainers" / "the retainers" (calque-ish paraphrase, thread held 2/2), bēah-wriða → "a ring-band" (calque; "armlet" and "arm-ring" were the substitutions available and would have been more idiomatic), lind-plega → "the shield-play" (calque; "battle" was the substitution), here-grīma → "his battle-mask" (calque; "helmet" was the substitution and is what the word denotes), wæl-stōw → "the ground of slaughter" (calque), wīg-bealu → "the evil of war" (calque; H&S glosses the phrase "to stir up strife", which was the substitution available), māððum → "the treasure", frætwe → "the war-gear", bill → "the blade" (see 3b).

So on this passage, the majority of the realia were forked, and I answered nearly every fork the same way — calque. That is a uniform answer, unlike the four-way handling I committed in the Genji passage, and I noticed the uniformity while doing it. It is not a principle I adopted; it is that Old English realia are overwhelmingly transparent compounds — here-grīma is "army-mask", lind-plega is "shield-play", wæl-stōw is "slaughter-place" — so the calque is always sitting right there, fully formed and fully legible, in a way that 三途の河 never is for English. The source's own word-formation supplies the handling.

(e) The proper names: free copy, and the opacity is not inherited.

Every name in the passage copies without cost: Hrothgar, Freawaru, Froda, Ingeld, the Scyldings, the Heathobards, Withergyld. In the Japanese cell that would have been the end of it. It is not, because Old English personal and tribal names are transparent common nouns, and the transparency does not survive. Frēa-waru is "lord-protection", and she is being sent as protection between two lords. Heaðo-beardan is "battle-beards". Wiðer-gyld is literally "counter-payment", requital — and the glossary of the edition I am using glosses it as the common noun first and marks the proper-name reading with a query.

That last one made me stop for a long time. H&S read 2052 syððan wiðer-gyld læg primarily as "after requital lay [low]"; the standard modern reading is that Withergyld is a dead Heathobard warrior. I wrote "when Withergyld lay dead", departing from my edition's primary gloss, for two reasons: the surrounding clauses are a sequence of concrete battle events (the Danes struck him down, held the field), and in modern English the common-noun rendering collides with an idiom — "when requital lay low" reads as keeping out of sight. The alternative I rejected, "when requital had fallen," is grammatical and dead.

The point for the study limb is what the copy costs. Wiðergyld in modern English is an opaque foreign-looking string, exactly as Sanzu-no-Kawa is in A-shaw-spider-thread. Intralingual copy manufactured a new opacity here, which is the thing the Genji anchor concluded intralingual copy never does. The difference is not the language boundary; it is whether the target's reader still holds the sense the signifier had. Logged here as an observation from inside the translation; the study limb tests it.

(f) Two textual cruxes, decided and flagged.

• 2030b Oft nō seldan hwǣr. H&S print a double litotes — "often, nowhere seldom" — where the manuscript and most modern editions have oft seldan hwǣr, one of the poem's well-known cruxes. I translated the text in front of me: "Time and again, and nowhere seldom." It is odd English. It is odd Old English too, and I did not smooth it.
• 2036 duguða biwenede. A past participle with a genitive plural, and bewenian is not in this edition's glossary at all. "Attended by the retainers" is my reading; "well entertained among the host" is equally available and would attach the phrase to the Danish party as a whole rather than to the young man. Chose the reading that keeps the singular focus of þonne hē … gǣð.

3. Smaller choices, logged because they were choices

• (a) No archaism, and this decision is partly primed. The whole thing could have been pitched in a Victorian-medievalist English. I wrote plain modern prose. Two reasons: the poem was not archaic for its audience, so archaic English translates the manuscript's age rather than the poem; and — declared as priming — the session's baton told me that one of the study limb's comparison texts (Morris & Wyatt) is heavily archaizing, and being the plain pole of a three-point scale is more useful to the study than being a second archaizer. I did not read Morris. But I knew what I was positioning against, and that is not a source-only decision.
• (b) Prose, not verse. No alliterative imitation was attempted. The source is strict alliterative verse with a medial caesura and I have reproduced none of it: gār-cwealm gumena, wīg-bealu weccean, blōd-fāg … bites all lose their sound-binding. This is the single largest formal loss in the translation and it is a whole-text loss, not a local one. Verse would also have been where my contamination is heaviest, since alliterative solutions to famous half-lines are the most memorable thing about published Beowulf translations.
• (c) bill → "the blade." billes bite at 2061. "Bill" survives in modern English only as the name of a specific polearm, so copying it would name the wrong weapon; "the sword's bite" is a substitution that names a weapon the line does not name. "The blade's bite" keeps the alliteration of the source half-line by accident, keeps the metaphor, and loses the weapon-word.
• (d) Threads I went looking for and held. sibb frames the digression — friðu-sibb folca at 2018, dryht-sibbe dǣl at 2069 — and I held it with "bond" at both ends ("the peace-bond of peoples", "the lordly bond"). dǣl 2/2 with "portion". wine 2/2 with "friend". sealde 2/2 with "gave", to keep the queen's giving and her daughter's giving as one gesture. duguð 2/2 with "retainers". eorl 2/2 with "nobles". hæleð 2/2 with "fighting men". The searching cost real time and most of these are invisible in the result.
• (e) A synonym set I could not reproduce. The source has secg, hæleð, guma, cempa, wiga, beorn-class words, þegn, eorl and dryht-bearn all available for "man / warrior", because alliterative verse needs a word beginning with any letter. Modern English has a much thinner set and no metrical reason to reach for it. I flattened: "a man", "the fighting men", "men", "the young warrior", "spear-fighter", "thane", "the nobles". The variation in my text is arbitrary where the source's is functional.
• (f) Word order and apposition. Old English apposition — a noun, then a second noun re-naming it, then a third — is the poem's basic rhetorical unit (eald æsc-wiga, sē þe eall geman gār-cwealm gumena). Modern English tolerates it only in short bursts before it reads as a list. I kept it where I could and used dashes to license it; at 2036 and 2053 the appositives (dryht-bearn Dena, hwate Scyldungas) had to be pinned with dashes and commas to keep them from reading as new subjects.
• (g) hyra sylfra feorh → "their own lives." Old English feorh is singular here, one life apiece; English pluralises. Silent.
• (h) Articles. Old English has no indefinite article and its definite demonstrative is optional. Modern English requires a determiner on nearly every count noun, so "a ring-band", "the ale-cup", "a ring", "the treasure" are all decisions the source did not have to make. This is a second, smaller forced explicitation, running under the whole translation rather than at named sites.

4. What I could not do

• The metre and the alliteration (3b). Absent, not reduced.
• The transparency of the names (2e). Copied and opaque.
• The kenning system as a system. Individual compounds are calqued and mostly survive; what does not survive is that they are generated — that a poet could make a new one at need and an audience would parse it instantly. My "shield-play" and "battle-mask" read as quaint fixed expressions rather than as live formations.
• duguð / geogoð as a social pair. duguð at 2021 and 2036 is one half of an institutional contrast — the proven warriors as against the young retainers — which the poem uses elsewhere. "Retainers" names a job. The rank structure inside the word is gone, the same loss as めざまし in the Genji passage.
• The litotes habit. Oft nō seldan hwǣr, nō on wealle leng, and the whole understating temper of heroic speech. I preserved the one at 2030 because it is conspicuous; the general tendency of the style, I did not attempt.

5. Two evaluative sentences, flagged

internal-judgment-only: (i) the line I am least confident about is 2036 duguða biwenede, where the verb is absent from my glossary and both available readings change who the phrase attaches to; (ii) I think "hall-floor" is the weakest thing in the translation as English, and I kept it anyway because dropping it would have cost the 2035/2055 repetition, which is the passage's sharpest formal feature.