N1 2015-07 — audit report

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

OCR confidence

Source: retypeset-digital — quality 5/5: Crisp retypeset digital text throughout; watermark headers on every page stay in the margins and never overlap body text. Reading-section pages use very small dense type but remain fully legible. Being a third-party retypeset, occasional transcription typos are possible (a few suspect characters in 問題3 stems), and no furigana is present.

Every section was transcribed twice independently (a full-page pass and a double-resolution half-page pass) and reconciled. Of 13 sections, 1 agreed on all content between the two passes (reconciled automatically, differing only in transcriber notes) and 12 had at least one content difference resolved by a third adjudication pass that re-read the page images. 128 leaf-level differences were examined in total.

Residual OCR risks (31):

• 問題1: Q5 contains a printed source typo ("淡いと"); transcribed as-is per conventions and flagged in notes, but the intended original wording is uncertain.
• 問題1: Furigana absent throughout this section (none printed), consistent with the dense retypeset layout.
• 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: Q18 stem is a genuine source garble ('錯覚のもしかたがない'); transcribed verbatim as required, so it intentionally reads ungrammatically.
• 問題3: Source is a third-party Chinese retypeset; other unflagged silent typos could exist but none were observed in mondai3 14-19 beyond the Q18 garble.
• 問題4: Source is a third-party Chinese retypeset; any retype typos are transcribed as printed and none were flagged as suspicious in this section.
• 問題4: No furigana present in this section, so no ruby disambiguation was required.
• 問題5: The duplicate boxed 28 (no 29) is a genuine source defect, not OCR error; downstream consumers expecting contiguous 26-35 numbering must handle the gap (final has two number:28 entries and no 29).
• 問題5: Q31 option 1 (出たせいなのかは), Q35 label 回中, and Q35 option 3 (見ていだだいた) are source-retypeset typos transcribed verbatim; they look wrong but match the print.
• 問題5: Q34 line-2 inter-speaker gap is rendered as 2 full-width spaces based on a ~140px measured gap; the exact intended count is inherently ambiguous for a proportional-ish gap, but 2 is the closest deterministic read.
• 問題6: The instruction-line ＿★＿ uses full-width low lines per convention; if the project prefers a different normalization for the star-inside-blank instruction case, this single field may need adjustment.
• 問題6: Slot-run leading/trailing single half-width spaces are present per convention but are visually impossible to fully verify against the printed underline widths; they follow the deterministic _ _ ★ _ pattern.
• 問題7: Q44 options 3 'これらまで' and 4 'あれらから' are almost certainly retype garbling of 'これまで'/'あれから', but are transcribed verbatim per convention.
• 問題7: Citation has no punctuation between '中島隆信' and '読売新聞' (printed solid); transcribed as printed — could be a source typo, not corrected.
• 問題7: Citation/注 print order on the page is citation-above-注; the convention also calls the citation 'the final <p>'. Resolved in favor of physical print order (citation then 注). If the project later standardizes citation-last, this single ordering may need a swap.
• 問題8: Passage (1) body lines sit at the very bottom of p05-b at small native resolution; resolved via 2x upscaled crops which were legible, but tiny kana (とか vs から, り vs nothing) carry slightly more uncertainty than the rest.
• 問題8: The source is a third-party retypeset; several printed typos (売れるり, 発堀, そかは, あるのでる, the stray comma in the 梅原猛 citation) are transcribed verbatim per convention and are not errors in this transcription.
• 問題9: Instruction line: the stray mark between 文 and 章 is small; rendered as 、 but could be a stray 。 — low impact, flagged in notes.
• 問題9: Passage (1) (注1): 強い思い、から逃げられない — the comma placement and 思い vs a possible 思い込み are at the resolution limit; kept as 思い、 per best reading.
• 問題9: Passage (1) citation: kept the full-width space inside 『本の読み方　スロー・リーディングの実践』; the gap is a real space in the image but its exact width is a judgment call.
• 問題9: Printed typos retained as-is: 国有 (passage 3 注3, for 固有) and the instruction stray comma; these are source retypeset artifacts, not OCR errors.
• 問題10: The N1 reading-passage convention normally uses the 2-em ―― for long dashes; here the print uses single em-dashes, so the dash rendering is image-confirmed but departs from the usual pattern — flagged in notes in case the house style wants normalization.
• 問題10: Source is a third-party Chinese retypeset; the okurigana inconsistencies (好い/よい, 分らない) are reproduced as printed and may not match the original published essay.
• 問題11: The author string 角屋重樹森本信也編著 genuinely has no visible separator between the two surnames in this retypeset; transcribed as printed, which may differ from the original publication.
• 問題11: Passage B's 訊ねれば uses 訊 (vs the more common 尋); the glyph is small but legible at double resolution and matches A.
• 問題11: All 10 content differences resolved to A; B's output is a paraphrase rather than a faithful OCR, so the only residual risk is a shared A/B error, mitigated by the five double-resolution spot checks above (none found).
• 問題12: The author kanji 眞 vs 真 was read at moderate zoom; both transcribers agree on 眞 and it is plausible for this author's name, so risk is low.
• 問題12: Q68 option 3 '可能' is almost certainly a source typo for '可能性'; transcribed verbatim per convention, but a downstream consumer might mistake it for an OCR error.
• 問題13: 実施回数：colon — read as full-width from a slightly wider gap, but at this print density half- vs full-width is hard to be 100% certain.
• 問題13: 受付時間 colon: both transcribers used half-width '受付時間:'; the image gap is ambiguous and could be full-width like 電話：; kept the agreed half-width value.
• 問題13: The half-width vs full-width distinction for the 留意点 spaces and the space before 場合 is judged from gap width at retypeset resolution and could plausibly be the opposite in a couple of spots.

Source-text quirks the OCR transcribed verbatim (printed typos/recall artifacts; the ones judged to be retype errors were corrected in the layer below) — 31 noted at OCR time:

• 問題1: Question 5 stem reads "庭に淡いとピンクの花が咲いている。" as printed; the "と" appears to be a typo in the source retypeset (likely should be "白い" or similar), transcribed as printed.
• 問題3: Q18 stem reads as printed: 「佐藤さんが錯覚のもしかたがない。」 — looks like a retypeset typo (likely intended 佐藤さんが錯覚するのもしかたがない or 錯覚を起こすのもしかたがない); transcribed as printed.
• 問題5: Question-number anomaly: the boxed numbers print 26, 27, 28, then 28 again (the 南北川 water-quality question), then 30, 31, 32, 33, 34, 35. There is no boxed 29; box 28 appears twice. Verified against p04-t.png; transcribed both as printed per the authoritative-print rule.
• 問題5: Q31 option 1 prints as「出たせいなのかは」which appears garbled/typoed in the source retypeset; transcribed as printed.
• 問題5: Q34: line 1 prints the wife's and husband's quotes adjacent with no intervening space (「…だって。」夫「…); line 2 prints a wide gap (~2 full-width spaces) between 妻「…待とう。」 and 夫「うん。」, transcribed as two full-width spaces. Verified by pixel measurement on p04-t.png.
• 問題5: Q35: the third speaker label is printed 回中 (likely a retype typo for 田中, the speaker in the first line); transcribed as printed.
• 問題5: Q35 option 3: printed 見ていだだいた (だだ where いただいた would be expected); transcribed as printed, suspected typo.
• 問題7: Q44 options 3 'これらまで' and 4 'あれらから' read literally as printed; they look like retype typos (plausibly 'これまで' / 'あれから'), but transcribed verbatim. Confirmed against double-resolution renders.
• 問題7: Citation prints publisher and date without punctuation between '中島隆信' and '読売新聞'; transcribed as printed: '中島隆信読売新聞 2013年3月24日付朝刊による'. Date is 3月24日 (verified against the high-resolution render).
• 問題8: Passage (1) citation prints 『半径1メートルの「売れるり発想術』 — an apparent retype typo (extra り in 売れる, and an opening 「 with no matching closing 」 before 』). Transcribed as printed.
• 問題8: Passage (1) paragraph 2: the three quoted terms are printed in corner brackets — 「売れる発想」「売れるシナリオ」「売れる商品」 — not underlined; the underline in the passage appears only via the question 46 reference. Transcribed as printed.
• 問題8: Passage (2): 発堀 is a likely retype typo for 発掘; transcribed as printed.
• 問題8: Passage (4): そかは (likely それは) and あるのでる (likely あるのである) appear to be retype typos; transcribed as printed. The citation 『饗宴―梅原猛随想、と対話』 has an unusual comma before と対話; transcribed as printed.
• 問題9: Source is the 日语轻松考 Chinese retypeset of N1_2015-07. The B transcription (worked from double-resolution half-page renders) matched the images closely across all three passages; the A transcription was heavily garbled in the passage bodies (it admitted being near the legibility limit), so the adjudicated passages follow B's reading verified against p07-t/p07-b/p08-t/p08-b/p09-t.
• 問題9: Passage (1) citation title prints 『本の読み方　スロー・リーディングの実践』 with a full-width space after 読み方 and a ・ in スロー・リーディング (A's reading); B dropped both — final keeps them as printed.
• 問題9: Passage (3) (注3) gloss prints 「人やものに備わる国有の性質」; 「国有」 is almost certainly a typo for 「固有」, transcribed as printed.
• 問題10: Passage uses inconsistent okurigana as printed: 「これが好い」 (kanji 好い) in one sentence but 「これがよい」 (kana) in the next; also 「分らない」 in paragraph 3 lacks the 分か okurigana while paragraph 1 uses 「分からない」. Transcribed as printed.
• 問題10: The parenthetical in paragraph 3 is set off by single em-dashes (one U+2015 each): 踏み出し―あるいは…しれない―は. Verified against the double-resolution image as a single ― glyph (one character width), not the 2-em ―― pattern.
• 問題11: Citation author string is printed as 「角屋重樹森本信也編著」 with no separator between the two names (角屋重樹 and 森本信也); transcribed as printed.
• 問題11: Source is a dense Chinese retypeset (日语轻松考); transcriber B systematically paraphrased rather than copied, dropping whole sentences from both passages and altering answer wording. Every content difference between A and B was adjudicated against the double-resolution p10-t/p10-b renders and resolved to A's reading, which matches the print exactly.
• 問題12: Q68 option 3 prints "自分が持つ可能が最大限に発揮できる" — "可能" appears to be a retype typo for "可能性" (the passage itself uses "自分が持つ可能性を大事にしたい"). Transcribed as printed.
• 問題12: The （注3）marker is printed in the passage body as "何かしらの（注3）変化を察知するにつけ" (verified on p11-t); transcriber B's doubt about this marker is unfounded.
• 問題12: Transcriber B's passage/options were heavily garbled (likely from the half-resolution crops); A's clean reading was confirmed verbatim against the double-resolution p10-b and p11-t images.
• 問題13: 留意点 third item prints 「選択式の箇所が末入力であった」 — 末 (sue) is almost certainly a retype typo for 未 (mi, 未入力); transcribed as printed.
• 問題13: 留意点 third item has a half-width space before 場合 (「…末入力であった 場合」) from a line break in the source; transcribed as printed.
• 問題13: Q70 option 4 prints 「すべでの質問」 — すべで is a typo for すべて; transcribed as printed.
• 問題13: 応募条件 (1) prints 「以下にお住まいの方関東に1都3県(…)」, awkward/garbled phrasing in the source; transcribed as printed.
• 問題13: 応募条件 (4) prints 「新聞、マスコミ、広告関係の職業従事されていない方」 — likely a dropped に before 従事; transcribed as printed. Only 「従事されていない方」 is underlined.
• 問題13: 謝礼 line prints 「任期終了時」; could be a typo for 期間, but transcribed as printed.
• 問題13: 実施回数 label uses a full-width colon (「実施回数：」) whereas the other labels (モニター契約期間:, 内容:, 謝礼: …) use half-width colons; transcribed as printed.
• 問題13: The 過去に line uses a small half-width asterisk (*), whereas the three 留意点 items use full-width ＊ marks each followed by a half-width space; transcribed as printed.

Corrections applied (20)

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

• 問3 q15 question: 「<u>互角</u>だ」 → 「<u>互角だ</u>」
  • The correct choice 大体同じだ is a complete predicate; leaving だ outside the tested span produces 大体同じだだ. The full target is 互角だ.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問3 q18 question: 「<u>錯覚</u>するのも」 → 「<u>錯覚する</u>のも」
  • The correct choice 勘違いする is a complete verb; leaving する outside the tested span produces 勘違いするする. The full target is 錯覚する.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問3 q19 question: 「<u>殺到</u>した理由」 → 「<u>殺到した</u>理由」
  • The choices are complete past-tense clauses, including the correct 一度に大勢来た. Leaving した outside the tested span makes their substitutions ungrammatical; the full target is 殺到した.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問1 q5 question: 「<u>淡い</u>とピンク」 → 「<u>淡い</u>ピンク」
  • The と after 淡い is a retype artifact; the natural sentence is 淡いピンクの花 (pale pink flowers). The intervening と is spurious and ungrammatical.
  • evidence: jlpt-freq witness (2015.07_2015-7真题.txt l.30): 庭に淡いピンクの花が咲いている。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問3 q18 question: 「<u>錯覚</u>のもしかた」 → 「<u>錯覚</u>するのもしかた」
  • Dropped する: the original reads 錯覚するのもしかたがない. Our text omits する, leaving the ungrammatical 錯覚のもしかたがない.
  • evidence: jlpt-freq witness (真题 l.79): そんなことを言われたら、佐藤さんが錯覚するのもしかたがない。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問5 q35 question: 「<p>回中「そうですか。」 → 「<p>田中「そうですか。」
  • 回中 is a retype typo for 田中, the same speaker as the first line of the dialogue (only two speakers, 田中 and 部長).
  • evidence: jlpt-freq witness (真题 l.190 and l.192): both 田中 lines; 田中「部長… and 田中「そうですか。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q44 passage: 「（中島隆信読売新聞」 → 「（中島隆信　読売新聞」
  • A separator (space) belongs between the author name and the publication name; the two are printed solid in our text.
  • evidence: jlpt-freq witness (真题 l.248): （中島隆信 読売新聞2013年3月24日付朝刊による） [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q46 passage: 「『半径1メートルの「売れるり発想術』」 → 「『半径1メートルの「売れる！」発想術』」
  • The book title is garbled: り is a retype error for ！」, which also supplies the missing closing bracket for the opening 「. Correct title is 半径1メートルの「売れる！」発想術.
  • evidence: jlpt-freq witness (真题 l.274): 『半径1メートルの「売れる！」発想術』 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q47 passage: 「重要なものを発堀して」 → 「重要なものを発掘して」
  • 発堀 is a non-word; 発掘 (excavate/unearth) is the intended word in this metaphor about libraries unearthing important information. Both available witnesses show OCR corruptions (発堀/発坦) of the same underlying word.
  • evidence: clear non-word; 発掘 is standard Japanese. Witnesses corrupt it (真题 l.283 発堀; n1-2015.7真题 l.268 発坦). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q49 passage: 「しかし、一つの眼である限り、そかは世界を歪んで」 → 「しかし、一つの眼である限り、それは世界を歪んで」
  • そかは is a retype typo for それは.
  • evidence: jlpt-freq witness (真题 l.318): 一つの眼である限り、それは世界を歪んで見ているのである [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q49 passage: 「見えない何かがあるのでる。」 → 「見えない何かがあるのである。」
  • あるのでる is a retype truncation of あるのである.
  • evidence: jlpt-freq witness (真题 l.318-319): その眼からはどうしても見えない何かがあるのである。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q50 passage: 「強い思い、から逃げられない」 → 「強い思いから逃げられない」
  • The comma between 思い and から is a spurious retype artifact; the gloss reads 強い思いから逃げられない.
  • evidence: jlpt-freq witness (真题 l.353): ここでは、強い思いから逃げられない。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q56 passage: 「人やものに備わる国有の性質」 → 「人やものに備わる固有の性質」
  • 国有 (state-owned) is nonsensical here; 固有 (inherent) is the intended word in the gloss of 属性.
  • evidence: jlpt-freq witness (真题 l.452): 属性：人やものに備わる固有の性質。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q68 answers[2]: 「自分が持つ可能が最大限に」 → 「自分が持つ可能性が最大限に」
  • Dropped 性: the option should read 自分が持つ可能性が最大限に発揮できる, matching the passage's 自分が持つ可能性. 可能 alone is a truncation.
  • evidence: jlpt-freq witness (真题 l.649): 自分が持つ可能性が最大限に発揮できるような目標 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「広告関係の職業<u>従事されていない方</u>」 → 「広告関係の職業に<u>従事されていない方</u>」
  • Dropped particle に before 従事; the original reads ...職業に従事されていない方.
  • evidence: jlpt-freq witness (真题 l.662): 新聞・マスコミ・広告関係の職業に従事されていない方 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「選択式の箇所が末入力であった」 → 「選択式の箇所が未入力であった」
  • 末 is a retype typo for 未; 未入力 (not yet entered) is the intended word.
  • evidence: jlpt-freq witness (真题 l.688): 選択式の箇所が未入力であった場合 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q70 answers[3]: 「すべでの質問に回答する。」 → 「すべての質問に回答する。」
  • すべで is a retype typo for すべて (で should be て).
  • evidence: jlpt-freq witness (真题 l.740): 回答形式を問わず、すべての質問に回答する。 [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問5 q35 answers[2]: 「見ていだだいたと思うんですが」 → 「見ていただいたと思うんですが」
  • いただいた is the standard honorific past form; だだ is a non-word retype slip for ただ (い-た-だ-い-た). Flagged in quirks note 10.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問7 q44 answers[2]: 「これらまで」 → 「これまで」
  • これらまで is not a real adverbial; options 1/2 are こうして/ああして (こ/あ demonstratives), so the distractor should be これまで. The spurious ら is a retype insertion. Quirks note 15.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問7 q44 answers[3]: 「あれらから」 → 「あれから」
  • あれらから is not a real form; the natural こ/あ-series distractor is あれから. Spurious ら inserted in retype. Quirks note 15.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]