N1 2016-07 — 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 4/5: Crisp retypeset digital text throughout; watermarks ('小程序：日语轻松考' header, '祝亲考级 通过~通过~通过~'/'Yuuki Bùi' corner marks) stay in margins and do not obscure content. Docked one point because several pages render at modest resolution (small option text and gloss notes require zoom) and the retypesetting drops original furigana/layout fidelity.

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. 96 leaf-level differences were examined in total.

Residual OCR risks (27):

• 問題1: No PIL/imagemagick/ffmpeg available to crop/zoom; underline-extent calls for Q2 and Q6 rely on the full-resolution p02-t.png render, which was legible and unambiguous.
• 問題1: Source is a Chinese retypeset; no typos were observed in this section but minor glyph fidelity cannot be 100% guaranteed.
• 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: Q14 option 2 prints 「部 分的に」 with a stray intra-word space; transcribed as 部分的に per convention, but the gap is genuinely present in the source.
• 問題3: No furigana present anywhere in the section (dense retype, as expected); none transcribed.
• 問題4: All six questions (20-25) and their underline spans were confirmed against the double-resolution halves; no structural or count discrepancies. Source is a Chinese retypeset, so any in-text wording oddities are transcribed as printed; none appeared suspicious in this section.
• 問題4: Half-width ASCII 'A社' (Q21/Q25) and 'A工場/A市' are printed full-width in the source but normalized to ASCII per conventions; both transcribers applied this consistently.
• 問題5: Q32 ends 問題5 cleanly into 問題6 at the bottom of p04-t; question count 26-35 (10 questions) confirmed, no merged/split anomalies.
• 問題6: Q40 occupies the densest row at the very top of p05; although the double-resolution half-page render is legible and was used to resolve all four options and the stem, this row remains the lowest-confidence area of the section.
• 問題6: The corrected '進めてはいるが' relies on a shared error in both transcribers; the high-res crop is clear, but it is worth a final human glance since neither independent pass caught it.
• 問題7: The citation mark after カツ丼 is rendered ambiguously between chōonpu (ー) and dash (―); read as chōonpu from glyph shape/position but the source is a low-res retypeset.
• 問題7: Passage prose was read from the standard-resolution top-half render (p05-t); individual interior kana were legible but the source is a dense retypeset and very small subscript-level marks carry residual uncertainty.
• 問題8: Passage (1) email body paragraphing (single <p> with <br> vs separate <p> per line) is a judgment call; chose one <br> block because line spacing is uniform with no blank-line gaps, but the source has no hard rule distinguishing letter lines from paragraphs.
• 問題8: 脹み / 過せる are genuine oddities of this Chinese retypeset (日语轻松考); preserved as printed per instructions, but they differ from the canonical published text and could be retype artifacts.
• 問題9: Passage (3) opening word: A read '今々の時代', B read '今の時代'. At double resolution the single glyph between 今 and の is at the legibility limit; adopted B ('今の時代'), which reads naturally and matches the published essay, but a printed 今々 cannot be fully ruled out.
• 問題9: Underlines in the passage bodies (①②③ spans in passages 2 and 3) are thin hairlines at this resolution; their presence is confirmed by the circled-reference markers but exact start/end pixels were judged, not measured.
• 問題9: Citation in passage (3) printed with spaces/hyphen as '2014 年6-7 月号'; transcribed as read (digits ASCII, hyphen half-width).
• 問題10: The source is a Chinese retypeset; transcribed as printed. No suspicious typos detected in the adjudicated text — every disputed string was resolved by direct image reading at double resolution.
• 問題10: Both transcribers had multiple independent paraphrase-style errors in paragraphs 2 and 4 (likely autocomplete/memory drift rather than OCR noise); the adjudicated passage was reconstructed strictly from the printed glyphs to avoid inheriting either side's drift.
• 問題11: The （注1）（注2）（注3） glosses physically reference passage A's terms but are printed in a shared row below passage B's box; placed at end of passageB to match print location — an alternative reading could attach them to passageA.
• 問題11: Q64 option 3 wording 'メンバーも意見を求めるべき' is grammatically odd (likely intended メンバーにも) but is a faithful transcription of the dense Chinese retypeset, not corrected.
• 問題11: Source is a third-party retypeset with no furigana; no ruby was present to verify, and any latent typos in the original retype are transcribed as-is.
• 問題12: The passage is a dense third-party retypeset; the adjudicated text was read carefully from 2x crops, but a few full-width spaces/exact punctuation in the long middle clauses could differ by a character.
• 問題12: The （注） vs （注1） label/number mismatch and the citation-before-注 ordering are transcribed as printed and are intentional, not errors to correct.
• 問題12: Author '佐倉純' is almost certainly a retype typo for the real author 佐倉統; left as printed per conventions.
• 問題13: Breadcrumb separator: printed glyph is a thin chevron; rendered as full-width ＞ (A). Could arguably be ASCII > but ＞ better matches the printed full-width chevron with surrounding spaces.
• 問題13: Three suspected source typos retained as printed (現在所→現住所, ご待参→ご持参, いただきまする→いただきます); all verified present in the hi-res image, so they are faithful to the retypeset 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) — 22 noted at OCR time:

• 問題5: Q27 option 2 printed as 「しつつも」; transcribed as printed.
• 問題6: Question 39 slot run starts immediately after を with no leading text gap; slot run placed between を and the following comma as printed.
• 問題6: Question 40 sits in a very dense top row of p05; verified against the double-resolution half-page render. Option 1 is '企業もまだ多い', option 2 'までに至らない', option 3 'と実感できる', option 4 'というのが', and the stem clause is '景気が回復してきている' (transcriber A's reading of this question's stem and options 1-3 was incorrect).
• 問題8: Passage (4) and Q49 options 3/4 use 脹み (脹みすぎて／脹みすぎたため) — kanji 脹 (U+8129) plus okurigana み, rather than the usual 膨らみ. Transcribed as printed; consistent across all three occurrences and confirmed in the high-res crop.
• 問題8: Passage (4) uses 過せる (printed without okurigana 過ごせる); transcribed as printed.
• 問題9: 問題9 covers questions 50-58 across three passages: (1)=50-52, (2)=53-55, (3)=56-58. All boxed question numbers verified against the images, matching the expected 50-58 range.
• 問題9: Passage (1) para 2: the （注1） reference marker is printed after 「これに尽きるん」 (i.e. 尽きるん（注1）じゃ), an unusual placement since the glossed term is 「これに尽きる」; transcribed as printed.
• 問題9: Passage (3) para 2 opening 「文字の言えない形で」 reads oddly grammatically; likely a retypeset oddity, transcribed as printed.
• 問題9: Dense multi-column retypeset layout; long passage paragraphs were cross-checked span-by-span against the double-resolution half-page renders.
• 問題10: Inline 注 markers (注1)–(注6) appear within the passage body and are transcribed as printed; the gloss block follows after the citation.
• 問題11: Passage A is printed as four indented paragraphs plus a source-citation line. The citation uses single corner brackets 「」 around the book title, transcribed as printed.
• 問題11: The （注1）（注2）（注3） glosses are printed as one three-per-line row below passage B's box (they reference terms フラットに/得てして/予定調和 that appear in passage A). Placed as three separate <p> at the end of passageB matching the print location; each marker is followed by a full-width space exactly as printed: （注1）　フラットに…
• 問題11: Passage A reads チームリーダーからメンバーへ質問 (not 一人一人へ) and そうやって出た結論 (not 出した). Passage B reads リーダーは会議でどうすべきか, 具体的方策 (no な), and 示し、メンバーと議論 (comma after 示し) — all verified against the double-resolution renders.
• 問題11: Q64 option 3: a stray justification space appears between 'べ' and 'きだと' in the dense retype ('求めるべ きだと'); removed as a line-wrap artifact, transcribed as the intended '求めるべきだと'. Wording 'メンバーも意見を求めるべきだ' reads slightly awkward but is transcribed as printed.
• 問題12: Transcriber B's passage reading was heavily corrupted/paraphrased across most paragraphs; the adjudicated passage follows transcriber A where A is accurate and the image where A also erred. Several A errors were corrected against the double-resolution scan: '現時点では適切と言うほかありません。「科学」ということばを使うか否かが大きな問題なのではありません。' (A had a different rendering), 'わたしたちはそれらの科学技術を使う主人公です' (A: 'まさにその'), the third underlined paragraph '③そのしくみなどを知る必要はない、という人もいるかもしれません。しかし、そのような使い方では、…「良い部分」しか聞かれないのだとしたら…。それらを使う主人公であるわたしたちは、…科学的背景やしくみを少しでも知った上で…' (A diverged from 'では'/'聞かれない'/'主人公'/'科学的背景やしくみ' onward), and 'つきあっていく、その第一歩は' (A: 'つきあって').
• 問題12: The reference mark in paragraph 2 is （注1） but the gloss at the foot is labeled just （注）; this label/number mismatch is transcribed as printed.
• 問題12: Citation author is printed as '佐倉純' (the real author of this work is 佐倉統) and the title is quoted with 「」 not 『』; both transcribed as printed.
• 問題13: Document headings use full-width square brackets ［］ (e.g. ［活動内容］), not lenticular 【】; transcribed as printed.
• 問題13: Suspected source typo: '現在所' in '＊面接時、学生証と現在所がわかるもの…' (expected 現住所). Transcribed as printed.
• 問題13: Suspected source typo: 'ご待参ください' in the ＊ line (expected ご持参ください, the form used correctly in ［応募方法］). Transcribed as printed.
• 問題13: Suspected source typo: 'ボランティア保険に加入していただきまする。' in ［その他］ (expected いただきます。). Transcribed as printed.
• 問題13: 謝礼（年額10,000）omits a currency unit/yen mark as printed; transcribed as printed.

Corrections applied (8)

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

• 問3 q14 question: 「<u>端的</u>に表している」 → 「<u>端的に</u>表している」
  • All four choices already end in に, so leaving に outside the tested span duplicates the particle. The full target is 端的に.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問13 q69 passage: 「ご待参ください」 → 「ご持参ください」
  • 待参 is a non-word; the correct 持参 is used elsewhere in the same document ((3)をご持参ください). Retype substituted 待 for 持.
  • evidence: Category homophone, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問13 q69 passage: 「いただきまする」 → 「いただきます」
  • まする is an archaic/incorrect form; standard polite is いただきます. The extra る is a retype artifact in this realistic document.
  • evidence: Category added-char, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問13 q69 passage: 「現在所がわかる」 → 「現住所がわかる」
  • 現在所 is a non-word; the intended word is 現住所 (current address), shown along with 学生証 at the interview. Retype substituted 在 for 住.
  • evidence: Category homophone, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q53 passage: 「組み合わせているということも、さらにあります」 → 「組み合わさっているということも、ざらにあります」
  • Subject is marked by が (あなたがした仕事が…組み合わさっている) requiring the intransitive 組み合わさっている, not transitive 組み合わせている; and the set phrase ざらにある (be common) fits, while さらにある is meaningless here — a dropped dakuten.
  • evidence: Category typo, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問11 q64 answers[2]: 「メンバーも意見を求めるべき」 → 「メンバーにも意見を求めるべき」
  • 意見を求める takes the person as に (メンバーに意見を求める); メンバーも is ungrammatical here. The に was dropped in the retype.
  • evidence: Category dropped-char, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問12 q65 passage: 「佐倉純／古田ゆかり」 → 「佐倉統／古田ゆかり」
  • The book おはようからおやすみまでの科学 is by 佐倉統 and 古田ゆかり; 純 is a retype error for 統 in the author credit.
  • evidence: Category referent, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q50 passage: 「これに尽きるん（注1）じゃないでしょうか」 → 「これに尽きる（注1）んじゃないでしょうか」
  • （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]