N1 2025-12 — audit report

66 questions (問題1–13, the non-listening half). Generated from the OCR and correction audit trails.

OCR confidence

Source: mixed — quality 3/5: Recall version (回忆版, attributed to 新干线/xgxgroup.cn, recirculated with a Xiaohongshu-style sticker on the cover) assembled from mixed sources: pages 1-25 are a clean retypeset recreation in the official JLPT layout (boxed question numbers, side tabs, furigana, vertical 問題8(4) box) that is crisp and fully legible, though pages 16/18/24 are low-resolution rasters (page 24 borderline-blurry) and page 5 degrades booklet pages 4-5 into a composite ending in bare plain text for Q25; pages 26-31 switch to genuine photocopy-grade scans of the real booklet with authentic 25-2-I-言・読-N headers, grainy but legible. Being a recall reconstruction, content fidelity is unverified and several recall/retype typos are visible (Q18 同様のこうが, Q25 目まくるしい, Q38 俳優のむ村川たかし, Q52 考え合う, Q59 技術て), plus one handwritten Chinese gloss inside the 問題7 passage. Legibility alone would merit a 4; fidelity risk and the degraded/composite pages bring it to 3.

Every section was transcribed twice independently (a full-page pass and a double-resolution half-page pass) and reconciled. Of 13 sections, 7 agreed on all content between the two passes (reconciled automatically, differing only in transcriber notes) and 6 had at least one content difference resolved by a third adjudication pass that re-read the page images. 83 leaf-level differences were examined in total.

Residual OCR risks (24):

• 問題1: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題2: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題3: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題4: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題5: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題6: The source is a community recall reconstruction, so even a perfect transcription (e.g. q38 俳優のむ村川たかし, q38 option 自然な演技は, q39 入賞し) may not match the real exam wording.
• 問題6: Slot-run blank widths vary in print (wrapped runs); per conventions they are normalized to four slot tokens around ★, so the printed visual widths are not preserved.
• 問題7: The handwritten-style scribble after 「クンクンが始まります。」 is illegible at available resolution; per convention it was not transcribed, but if it is actually typeset content (injected branding) a future higher-resolution pass could try to decode it.
• 問題7: 「お互いにおいをかぐ」 (single に) and 「敵が味方か」 are themselves plausibly recall typos in the source retype; they are transcribed as printed, but anyone using this text for grammar study should be aware the source is a community recall reconstruction.
• 問題7: No source citation is printed for the passage, so the passage cannot be cross-checked against a published original.
• 問題8: The instruction-line (1)/(4) glyphs were rendered as half-width ASCII parentheses to match project precedent and the digit-normalization rule; if the canonical reading should treat them as full-width Japanese parentheses, this single field would differ.
• 問題8: 特定的音 (q46 option 3) is preserved as a likely recall typo of 特定の音; intentional per the transcribe-as-printed rule.
• 問題8: 注 superscript markers are small in the full-page render; confirmed present in the double-resolution halves, so confidence is high.
• 問題9: （注4） subscript in p22-t prints at the far right of its line (near いた) due to line-wrap; placement after いそしんで is inferred from the gloss list (いそしむ) and the convention both transcribers applied to （注5）/（注6）, not from raw subscript x-position. Low risk but worth flagging.
• 問題9: Several recall typos transcribed as printed (q49 か, q52 考え合う, その意味で。, 最後的最後の階段, 認識できる contradiction) per the source being a community 回忆版 retype; not corrected by design.
• 問題9: Pages 22-23 (passage 4) are the official-layout retype, not booklet photocopies, so OCR confidence is high; no watermark/branding injection observed within problem 9 content.
• 問題10: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題11: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題12: The （注1） superscript on the blurry photocopy sits at the 想起／させる boundary; placement after 想起 is judged correct from the gloss target (想起する) and the parallel （注2）/コモディティ化 case, but the exact glyph boundary is mildly ambiguous in the scan.
• 問題12: Source-citation line is absent on the photocopy; it may have existed in the original booklet but been cropped/omitted — cannot be recovered from these pages.
• 問題12: 生活上役に立つ is unusual phrasing; transcribed as printed but cannot be 100% disambiguated from 生活の on the blurry copy (both transcribers and the 2x crop read 生活上).
• 問題13: Marker-spacing decision (no space after ■/①②③) is a judgment call: the gap is visibly present in print, and a reviewer favoring literal-gap representation could prefer B's full-width spaces. Chosen for internal consistency with the digit/address spacing-drop the conventions already mandate.
• 問題13: th-vs-td for the row-label cells is a semantic judgment; the photocopy gives no visual styling cue, so the choice rests on grid role. If the project later prefers <td> for non-styled label cells, the 送料/日数 row labels would flip to <td>.
• 問題13: This is a 回忆版 recall reconstruction; the address, prices, and all wording are as printed and may contain recall-introduced typos that cannot be verified against an official source.

Source-text quirks the OCR transcribed verbatim (printed typos/recall artifacts; the ones judged to be retype errors were corrected in the layer below) — 56 noted at OCR time:

• 問題1: q5: the source prints 芳しいない (with the underline covering 芳しいない) — almost certainly a recall typo for 芳しくない; transcribed as printed.
• 問題1: q5: printed as 「芳しいない」 (underline covering 芳しいない) — almost certainly a recall/retype typo for 「芳しくない」; all four options end in 〜しくない. Transcribed as printed.
• 問題2: q9: printed 「でコンクールでとちらが優勝するか」 — 「とちらが」 lacks the dakuten (suspected recall typo for 「どちらが」); transcribed as printed (verified at 2x zoom).
• 問題2: q9: no comma or particle between （　　）で and コンクール — printed as one run-on clause; transcribed as printed.
• 問題2: q11: printed 「大会本番に臨むだ。」 — suspected recall typo for 「臨んだ」; transcribed as printed (verified at 2x zoom).
• 問題2: q9: printed as 「とちら」 (no dakuten) where どちら is expected — verified at high zoom; suspected recall typo, transcribed as printed.
• 問題2: q11: printed as 「臨むだ」 where 臨んだ is expected — verified at high zoom; suspected recall typo, transcribed as printed.
• 問題2: q11: stem reads 「（　　）準備して」 without を after 準備, slightly unnatural; transcribed as printed.
• 問題2: q8: stem has double は (「試合の前半はAチームは1点…」), possibly a recall artifact; transcribed as printed.
• 問題3: q18 stem reads 同様のこうが as printed — likely a recall typo for 同様のことが (known source typo); transcribed exactly as printed.
• 問題3: q18: stem reads 「同様のこうが起きることは」 — 「こう」 is a suspected recall typo (likely for 事故/こと or similar); transcribed exactly as printed.
• 問題4: Question 25 is not in official format: its number is printed as "25." (not a boxed number), the headword is not underlined in any of the four options, and the options have no sentence-final periods. Transcribed exactly as printed, without adding underlines.
• 問題4: Question 25 headword is printed 目まくるしい — a suspected recall typo for 目まぐるしい (known source typo). All options repeat the まくるしい spelling.
• 問題4: Question 25 option 3 reads 日まくるしい (日 instead of 目) — transcribed as printed.
• 問題4: Question 25 option 4 begins 微夜で — suspected recall typo for 徹夜で; transcribed as printed.
• 問題4: Question 25 option 2 reads 新しいものが目まくるしい入れ替わっている (まくるしい directly before 入れ替わって, where まくるしく would be expected) — transcribed as printed.
• 問題4: q25 is printed as bare plain text below the booklet-style page (page p05 composites two source pages): the question number appears as "25." in plain text rather than the boxed style used for q20-q24, the options have no underlines under the headword, and the option sentences have no ending punctuation. Transcribed exactly as printed.
• 問題4: q25 headword printed as 目まくるしい — almost certainly a recall typo for 目まぐるしい (known source typo); transcribed as printed.
• 問題4: q25 option 3 prints 日まくるしい (日, not 目) — likely a further typo in the source; transcribed as printed.
• 問題4: q25 option 4 begins 微夜で — likely a recall typo for 徹夜で; the printed glyph reads as 微; transcribed as printed.
• 問題4: q22: the underline in option 3 covers only かすれ (before 、), while in options 1, 2, and 4 it covers かすれて; verified at high zoom.
• 問題4: Question 25: the headword occurrence in each option (printed without underline in this bare recall fragment) is wrapped in <u> to match the 問題4 target-marking convention; the recall typos (目まくるしい for 目まぐるしい, 日 in option 3, 微夜 in option 4) are preserved.
• 問題5: q27 answer 1 prints をはじめて (where the conventional expression would be をはじめ); transcribed as printed — likely a recall/retype artifact, though it is a distractor either way.
• 問題5: q27 option 1 is printed as をはじめて, which looks like a possible recall typo for をはじめ (or a distorted をおいて); transcribed as printed.
• 問題5: q32 stem reads こんな道が渋滞していなければ, where こんなに道が would be more natural; transcribed as printed.
• 問題6: q36: sentence ends 「も課長だ。」 as printed; with the options it forms the standard 〜も〜なら、〜も〜だ pattern (西山さんも西山さんなら、課長も課長だ), so the repetition of 西山さん/課長 in the options is intentional, not a typo.
• 問題6: q38: the stem prints 「俳優のむ村川たかしが」 — the extra む between の and 村川 is almost certainly a recall/retype typo for 「俳優の村川たかし」; transcribed as printed.
• 問題6: q38: option 1 is printed 「自然な演技は」 with は, where を might be expected before 見せた; transcribed as printed.
• 問題6: q39: stem reads 「マラソン大会で入賞し」 directly before the slot run; with option 2 ことが this suggests the source dropped a た (入賞した), but it is printed as 入賞し; transcribed as printed.
• 問題7: Suspected recall typo in paragraph 1: 「指先を出かすと」 (likely intended 「出すと」) — transcribed as printed.
• 問題7: Suspected recall typo in paragraph 2: 「ネコたちにとつでは」 with a full-size つ and で (likely intended 「にとっては」) — transcribed as printed.
• 問題7: Suspected recall typo in paragraph 2: 「敵が味方かなどの情報」 (likely intended 「敵か味方か」) — transcribed as printed.
• 問題7: Paragraph 3 uses hiragana 「ねこ」 (「ねこは肉食のハンター」) while the rest of the passage uses katakana ネコ — as printed.
• 問題7: Suspected recall typo in final paragraph: 「そのと触れても」 (likely intended 「そっと触れても」) — transcribed as printed.
• 問題7: q42 options 3 and 4 read 「示すてばかります」 and 「示すたものです」, both ungrammatical (likely recall typos for 「示してばかりいます」/「示したものです」 or similar) — transcribed as printed.
• 問題8: Passage (2) q46 option 3 reads 「特定的音が知覚できなくなること」 — likely a recall typo for 特定の音, but transcribed as printed.
• 問題9: Passage (1), q49 option 2 ends 「活発になったか」 (か, not から) where every other option ends in から; suspected recall typo, transcribed as printed.
• 問題9: Passage (2), q52 stem reads 「筆者の考え合うのはどれか」 — likely a recall typo for 「筆者の考えに合うのはどれか」 (missing に); transcribed as printed.
• 問題9: Passage (2): 「その意味で。「やりたいこと」…」 ends the clause with a full stop 。 where a comma 、 would be expected; transcribed as printed.
• 問題9: Passage (2): 「最後的最後の階段での賭け」 appears garbled — likely 「最後の最後の段階」 (的 for の, 階段 for 段階); transcribed as printed.
• 問題9: Passage (3): 「魚の体表の明暗の変化は十分に認識できることが明らかになっている」 is semantically contradictory with the preceding 「が」 (likely should be 認識できない); transcribed as printed.
• 問題9: Half-width digits/Latin normalized per convention: 「1〜35m」「30〜40cm」「2cm」「2018年」; surrounding typographic spaces dropped.
• 問題10: Suspected recall typo in q58 stem: 「…優先順位を低く見直す何のためか。」 — reads as if のは is missing after 見直す (expected 「見直すのは何のためか」). Transcribed as printed.
• 問題10: Suspected recall typo in q59 option 1: 「技術て人間の記憶力の優劣を測る」 — 技術て instead of 技術で (matches the known recall-typo list). Transcribed as printed.
• 問題10: q57 and q59 stems end in 「どう考えている。」/「どう述べている。」 without the usual final か; transcribed as printed.
• 問題10: q59 stem says 試験制度 while the passage's corresponding paragraph says 受験制度; possibly a recall inconsistency, transcribed as printed.
• 問題10: q57 option 4 uses よりよく in kana while the passage prints より良く; transcribed as printed.
• 問題10: q58 stem reads 「その案件の重みづけを変え、優先順位を低く見直す何のためか。」 as printed — likely a recall typo with とあるが/のは missing between the underlined quote and 何のためか; transcribed exactly.
• 問題10: q59 option 1 contains 「技術て人間の記憶力の優劣を測る」 — the known recall typo 技術て (for 技術で); confirmed at high zoom and transcribed as printed. (Task brief listed this typo under q59; it indeed appears in q59, not q59's stem.)
• 問題10: q57 stem ends 「どう考えている。」 and q59 stem ends 「どう述べている。」 without the usual final か — transcribed as printed.
• 問題12: 問題12 is a single passage with exactly questions 62-64 (no （1）（2） passage labels), matching the new 66-question format; boxed numbers 62, 63, 64 as printed.
• 問題12: Fifth paragraph reads 商品を生活上役に立つ存在だと認識してもらい — 生活上役に立つ is slightly unusual phrasing (not 生活の) but clearly printed in the genuine photocopy; transcribed as printed.
• 問題12: The passage repeatedly uses この様な (kanji 様) rather than このような; transcribed as printed.
• 問題13: Section headings are marked with a small filled-square glyph, transcribed as ■, and circled digits ①②③ appear in ご利用の流れ. Both markers show a visible gap before the following text, but that gap is treated as list-marker typographic spacing (consistent with dropping the digit/address spacing elsewhere) and no space is inserted after the glyph; ・ and ※ lines likewise sit tight against their text.
• 問題13: The fictional contact address is printed as 〒149-8547 東京都横山市村北1-9-13 (nonexistent address, as printed); gap after 「ピース」 is one full-width space, half-width space kept between 〒149-8547 and 東京都, and the small gap before 1-9-13 is digit glyph spacing, not a literal space.
• 問題13: Typographic half-width spaces around half-width digits in the photocopy (e.g. 合計 4,000円, 30%引いた, 送料 2,000円, 8 日後) were dropped as typesetting artifacts per the digit-normalization convention.

Corrections applied (32)

Evidence-based reconstructions of the original exam text. This section is the canonical correction log for the sitting.

• 問1 q5 question: 「芳しいない」 → 「芳しくない」
  • 芳しいない is malformed; the 問題1 reading options are all …しくない (3 かんばしくない is the answer), so the underlined word is 芳しくない. The 回忆版 source itself mistyped く as い. One unambiguous intended word.
  • evidence: vision-vs-scan p2 + reading-option constraint (かんばしくない)
• 問4 q25 question: 「目まくるしい」 → 「目まぐるしい」
  • 問題4 headword: the answer key confirms the tested word is 目まぐるしい; the 回忆版 mis-rendered it 目まくるしい (missing dakuten). Per Domenic's human judgement (OCR-confidence audit notes), implemented 2026-06-17.
  • evidence: Answer key + human judgement; the q25 options were printed in a low-quality rendering.
• 問4 q25 answers[0]: 「水が<u>目まくるしい</u>流れ」 → 「水が<u>目まぐるしい</u>流れ。」
  • Correct the headword (目まくるしい→目まぐるしい) and add the missing sentence period. Per Domenic's human judgement (OCR-confidence audit notes), implemented 2026-06-17.
  • evidence: Answer key + human judgement.
• 問4 q25 answers[1]: 「新しいものが<u>目まくるしい</u>入れ替わっている」 → 「新しいものが<u>目まぐるしい</u>入れ替わっている。」
  • Correct headword and add period. Per Domenic's human judgement (OCR-confidence audit notes), implemented 2026-06-17.
  • evidence: Answer key + human judgement.
• 問4 q25 answers[2]: 「夏の太陽が<u>日まくるしい</u>」 → 「夏の太陽が<u>目まぐるしい</u>。」
  • Correct headword (日まくるしい→目まぐるしい — also fixing 日→目) and add period. Per Domenic's human judgement (OCR-confidence audit notes), implemented 2026-06-17.
  • evidence: Answer key + human judgement.
• 問4 q25 answers[3]: 「微夜で今日はとても<u>目まくるしい</u>」 → 「微夜で今日はとても<u>目まぐるしい</u>。」
  • Correct headword and add period. Per Domenic's human judgement (OCR-confidence audit notes), implemented 2026-06-17.
  • evidence: Answer key + human judgement.
• 問2 q9 question: 「とちら」 → 「どちら」
  • とちら is a non-word; grammar requires どちらが優勝するか. Witness PDF and quirks log both show the dakuten-less とちら as a community-recall artifact, and the fix restores the only grammatical reading.
  • evidence: clear non-word (missing dakuten); grammar requires どちらが優勝するか; witness PDF line 79 reproduces とちら [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問2 q11 question: 「臨むだ」 → 「臨んだ」
  • 臨む is a godan verb whose plain past is 臨んだ; 臨むだ is ungrammatical. Sentence is 大会本番に臨んだ。
  • evidence: grammar: 臨む→臨んだ (godan past); witness PDF line 89 reproduces 臨むだ [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問3 q18 question: 「同様のこうが」 → 「同様のことが」
  • こう is a non-word in 同様の___が起きる; こと is the minimal single-kana fix giving 同様のことが起きる. Some ambiguity remains (事故 also fits), so confidence is downgraded, but the as-printed text is definitely a non-word and こと is the most natural minimal reconstruction.
  • evidence: clear non-word; grammar requires a noun; witness PDF line 130 reproduces 同様のこう (quirks log flags こと/事故) [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問6 q38 question: 「俳優のむ村川」 → 「俳優の村川」
  • The む between の and 村川 is a stray retype artifact; intended is 俳優の村川たかし. The tiengnhathay answer-key reconstruction shows 彼ならでは with no stray character.
  • evidence: clear stray character; answer-key-3 shows the assembled q38 sentence with no む [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問6 q39 question: 「マラソン大会で入賞し _」 → 「マラソン大会で入賞した _」
  • With the four options the sentence assembles as 入賞したことがよほどうれしかったのか、家に帰ってくるなり…; the た was dropped in recall (入賞しことが is ungrammatical).
  • evidence: answer-key-3 explicitly prints 大会で入賞したことが…(2413); grammar requires 入賞した [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q41 passage: 「指先を出かすと」 → 「指先を出すと」
  • 出かす is not a valid verb here; the intended phrase is 指先を出すと.
  • evidence: clear non-word; grammar (指先を出す); witness PDF line 318 reproduces 出かすと [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q41 passage: 「ネコたちにとつでは」 → 「ネコたちにとっては」
  • にとつでは uses a full-size つ where the idiom is にとっては.
  • evidence: clear non-word; idiom にとっては; witness PDF line 320 reproduces にとつでは [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q41 passage: 「そして敵が味方かなどの情報」 → 「そして敵か味方かなどの情報」
  • The 'whether A or B' construction requires 敵か味方か, not 敵が味方か.
  • evidence: grammar: alternative-question requires …か…か; witness PDF line 324 reproduces 敵が味方か [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q41 passage: 「そのと触れても」 → 「そっと触れても」
  • そのと is a non-word; the intended adverb is そっと (touch gently).
  • evidence: clear non-word; context (そっと触れても嫌がらない); witness PDF line 334 reproduces そのと触れても [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q46 answers[2]: 「特定的音」 → 「特定の音」
  • 特定的音 is ungrammatical (的音); the intended phrase is 特定の音, matching the passage's discussion of 音の知覚. This is a 問題8 reading option, not a deliberate 問題4 usage distractor.
  • evidence: clear wrong character (的 for の); 特定的音 is a non-word; witness PDF line 400 reproduces 特定的音 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q49 answers[1]: 「脳の活動が活発になったか」 → 「脳の活動が活発になったから」
  • The question is なぜか and every other option ends in から; this option drops ら, leaving か, which cannot express a reason.
  • evidence: clear dropped character; parallel options all end in から; witness PDF line 470 reproduces 活発になったか [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q52 question: 「筆者の考え合うのは」 → 「筆者の考えに合うのは」
  • 考え合う is not the intended form; the standard stem is 筆者の考えに合うのはどれか. The particle に was dropped.
  • evidence: grammar: 考えに合う; witness PDF line 518 reproduces 筆者の考え合う [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q51 passage: 「その意味で。「やりたいこと」」 → 「その意味で、「やりたいこと」」
  • A full stop 。 appears mid-sentence; the clause continues with …は、一種の「賭け」なのである, so a comma is required.
  • evidence: punctuation error; witness PDF line 495 reproduces the stray 。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q51 passage: 「最後的最後の階段での賭け」 → 「最後の最後の段階での賭け」
  • Garbled: 的 for の and 階段 for 段階. Intended is 最後の最後の段階での賭け.
  • evidence: wrong characters (的→の, 階段↔段階); witness PDF line 499 reproduces 最後的最後の階段 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q53 passage: 「相手の魚の色や輪郭ははっきりと認識できるが、魚の体表の明暗の変化は十分に認識できる」 → 「相手の魚の色や輪郭ははっきりと認識できないが、魚の体表の明暗の変化は十分に認識できる」
  • The contrastive が requires opposing clauses; as printed both are affirmative, which is self-contradictory. The shrimp is described as low-vision (0.3-equivalent contrast with the sharp-eyed fish), and the confirmed q53 answer is that the shrimp CAN perceive 明暗, implying it CANNOT clearly perceive color/outline.
  • evidence: answer-key-3 confirms q53 correct answer = shrimp can recognize 明暗; contrastive が; in-passage logic [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q57 question: 「筆者はどう考えている。」 → 「筆者はどう考えているか。」
  • JLPT question stems end with the interrogative か; this stem drops it before 。 (cf. q60/q64 which retain か).
  • evidence: convention: question stems end in か; witness PDF line 658 reproduces the missing か [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q57 answers[0]: 「人間は完成ではないので」 → 「人間は完全ではないので」
  • 完成(completion) is a kanji recall error for 完全(perfect); the intended sense is '(because) humans are not perfect', matching the passage's discussion of 完璧/不完全.
  • evidence: wrong kanji (完成 for 完全); witness PDF line 659 reproduces 完成 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q58 question: 「見直す</u>何のためか。」 → 「見直す</u>のは何のためか。」
  • Standard JLPT phrasing links the underlined quote to 何のためか via のは: 見直すのは何のためか。 The のは was dropped in recall.
  • evidence: convention/grammar: …のは何のためか; witness/quirks log records the gap after 見直す [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q59 question: 「筆者はどう述べている。」 → 「筆者はどう述べているか。」
  • Question stem drops the interrogative か before 。 (cf. q60/q64 retaining か).
  • evidence: convention: question stems end in か; witness PDF line 668 reproduces the missing か [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q59 answers[0]: 「技術て人間の記憶力」 → 「技術で人間の記憶力」
  • 技術て is ungrammatical; the particle is で (技術で…測る). て-for-で is a known recall typo in this sitting.
  • evidence: wrong particle (て for で); witness PDF line 669 reproduces 技術て [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問4 q25 answers[1]: 「新しいものが<u>目まぐるしい</u>入れ替わって」 → 「新しいものが<u>目まぐるしく</u>入れ替わって」
  • This is the correct answer for q25 (correctAnswerIndex 1) and must be grammatical: 目まぐるしく入れ替わる ('change rapidly') requires the adverbial 〜く form before the verb 入れ替わっている, not the attributive/predicate 目まぐるしい. The whole q25 is a bare recall fragment carrying the 目まくるしい→目まぐるしい retype lineage, and the adverbial inflection was lost in the same garbling.
  • evidence: Category typo, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問4 q25 answers[3]: 「微夜で今日はとても」 → 「徹夜で今日はとても」
  • 微夜 is not a Japanese word; the gloss/context ('今日はとても目まぐるしい') fits 徹夜で ('having pulled an all-nighter'). 微 and 徹 are visually similar; this is a separate retype typo from the target word, so it is fixable even though the option is a distractor.
  • evidence: Category typo, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問7 q41 passage: 「近づけてお互いにおいをかぐのが」 → 「近づけてお互いににおいをかぐのが」
  • Intended reading is お互いに+においをかぐ ('sniff each other's smell'), which requires two に sounds (the に of お互いに and the に of におい). The printed text has only one に (お互いにおいを), leaving おい as a non-word. One に was dropped; restoring it gives お互いに においをかぐ.
  • evidence: Category dropped-char, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q0 passage: 「（注1）するク」 → 「する（注1）ク」
  • （注N）note marker repositioned to follow its glossed term (the retype placed the marker before the word).
  • evidence: Marker-only move; the marker-stripped text is byte-identical. [note-marker relocation pass, 2026-06-20]
• 問6 q36 question/correctOrder: 「自分の主張を譲らない西山さん _ _ ★ _ も課長だ。; correctOrder null」 → 「自分の主張を譲らない西山さん _ ★ _ _ も課長だ。; correctOrder [1,2,0,3]」
  • The accepted assembly is 西山さんも西山さんなら、課長も課長だ, order 2314. Since the answer key gives q36 answer 3, ★ belongs in the second slot.
  • evidence: aixinjp 2025-12 answer/order summary gives q36 自分の主張を譲らない西山さん2も3西山さん1なら4課長も課長だよ; see http://aixinjp.com/a/lianxifangshi/zhentidaan/2025/1208/1243.html. [order-conflict cleanup, 2026-06-24]
• 問6 q37 question/answers[2]/correctOrder: 「トラブルが _ _ ★ _ トラブルだ。; answers[2]=起きてしまう; correctOrder null」 → 「トラブルが _ ★ _ _ トラブルだ。; answers[2]=起きてしまうのが; correctOrder [0,3,1,2]」
  • The accepted assembly is トラブルが起きないようにどんなに気をつけていても起きる時に起きてしまうのがトラブルだ, order 1423. Since the answer key gives q37 answer 4, ★ belongs in the second slot; のが was dropped from option 3 in the local retype.
  • evidence: Jiaoyubao 2025-12 answer/order summary gives q37 order 1423 and the assembled option text 起きてしまうのが; see https://www.jiaoyubao.cn/news/n268576.html. [order-conflict cleanup, 2026-06-24]

Flagged unresolved text issues (1)

Ambiguous source-text issues left unchanged rather than guessed.

• 問1 q2 question — reverted-overzealous-underline-trim
  • A prior pass trimmed the printed <u>潜んで</u> to <u>潜ん</u>, claiming the で fell outside the underline. Reverted (correction removed): the 問題1 reading options all end in んで (4 ひそんで is the answer), so the read/underlined word is 潜んで, not 潜ん. The full 潜んで stays underlined, matching the scan (a rendered source page (p02-t)).