N1 2014-12 — 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 3/5: Clean, crisp Word-retype with no watermarks, fully legible throughout, plus a rare bonus answer key for the written half. Marked down because the retype discards structure: answer options are unnumbered bullets (a mismatch hazard against the 1-4 answer key), document item numbers restart at 問題5, 問題8, and the listening section instead of using the official boxed 1-70, the 問題13 intro sentence describes a credit-card brochure while the actual task is the 就職支援行事 schedule (text pasted from another exam), and the two non-retyped elements (ビタミン生活V notice on p9, schedule table on p18) are small, fuzzy embedded scans. Type is dense and small on pages 7-18.

Every section was transcribed twice independently (a full-page pass and a double-resolution half-page pass) and reconciled. Of 13 sections, 3 agreed on all content between the two passes (reconciled automatically, differing only in transcriber notes) and 10 had at least one content difference resolved by a third adjudication pass that re-read the page images. 99 leaf-level differences were examined in total.

Residual OCR risks (29):

• 問題1: Q6 underline extent ('上拒んで') is an apparent source error; transcribed faithfully and flagged, but a downstream consumer expecting the headword 拒んで under the underline should be aware the span as printed includes 以上's 上.
• 問題1: Instruction-line blank: the actual printed glyph run is a plain underline ('______'), not literal full-width low lines; normalized to ＿＿＿ per convention rather than preserving the underline visual.
• 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: p04-t (double-res top half) renders small; Q22-24 readings were confirmed against the clear full-page p04.png plus p04-t, but a few kana are at the limit of legibility. Q24 opt2 'もてなしたい' vs 'もてなそう' tail is the least crisp but reads as 'もてなしたい'.
• 問題4: Underline span choice for inflected verbs (携わって) is a convention call, not visible in the source (source shows no underlines); applied consistently with 抱え込ま/抱え込ん/携わり.
• 問題4: The headword 抱え込む / 裏腹 / etc. on the 20.問題4 line and subsequent headword lines are not underlined in the source; underlining is editorial per convention.
• 問題5: Task brief expected 問題5 to be questions 3-35 (33 items), but this print only contains items 1-10 for 問題5; if a different source/edition were intended the scope would differ. Both transcribers independently read the same 10-item print, so confidence is high that the print is correctly captured.
• 問題5: The fill-in blanks are printed with half-width-looking parentheses '( )' in the source; per convention they are normalized to full-width （　　）. This is a deliberate normalization, not an OCR uncertainty.
• 問題5: Pages p06/p24 and their conventions (問題7 cloze, 問題13 table, embedded scans) are not part of this 問題5 section (covered by p04-p05 only); not applicable here.
• 問題6: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題7: The half-width ! before 」 in ではないか!」 is small in the scan; read as a half-width exclamation mark, but its exact width (half vs full) is mildly uncertain. Transcribed half-width per the digits/Latin normalization spirit, though punctuation width is normally preserved.
• 問題7: Inner-quote bracket usage is inconsistent in the source (「モーいいか」 vs 『モーいいか』 vs 『モーモー』). This is transcribed as printed and verified, but is the kind of inconsistency that could be a retype artifact.
• 問題7: The dash in the citation is rendered ―― (two U+2015) per convention; the single printed glyph is a long continuous rule, so the two-character rendering is a convention choice rather than a literal glyph count.
• 問題8: The (2) notice is a faded embedded scan; while p09-t/p09-b are legible, a few characters in the body (e.g. small kana in '弊社直営店舗') rely on contextual reading. The school-printed lead-in itself is crisp.
• 問題8: Passage (3) 'ことしたら' is an apparent retype typo; transcribed as printed per conventions, but the extra こ is faint and could conceivably be a rendering artifact.
• 問題8: 'お客様担当者' vs '担当：' in the scanned phone line is read as 'お客様担当者：0120-333-4455' from a slightly blurred region.
• 問題9: Inline reference-mark parens （注1）（注2） and some （中略） render narrower than gloss-line parens in the source; they are normalized to full-width per the convention's punctuation rule, but the page may have literally used half-width glyphs there.
• 問題9: Several options/stems contain genuine retype oddities (Q54 opt2, Q55 opt3, Q57 stem-vs-options mismatch) transcribed verbatim; they read awkwardly but match the page.
• 問題9: '――' in the citation 『人間自身――考えることに終わりなく』 is transcribed as two U+2015 per convention; the printed dash length was taken to be a 2-em dash.
• 問題10: The source is a Chinese prep-school retypeset; further latent retype typos elsewhere in the passage cannot be fully excluded, though every segment I spot-checked matched the scan.
• 問題10: 確かめる is read from a small-but-legible high-res glyph; the 確 radical is clear, but it is a low-frequency word choice in context (likely faithful to the retypeset rather than the original 岡本太郎 text).
• 問題11: Q64 four long answer options were identical between A and B and were not re-verified glyph-by-glyph against the image; they read naturally and a quick visual pass on p16-t showed no anomaly, but a residual shared-error risk remains.
• 問題11: Passage B 「肩書きだけで」: the crop is unambiguous, but this wording is slightly unusual in context (cf. A right / B 「に頼って」); transcribed as printed per the no-correction rule.
• 問題12: The passage is dense vertical→horizontal re-typeset prose; while every flagged diff was image-verified, a low risk of a shared, un-diffed correlated error remains in long agreed runs that were not exhaustively re-read.
• 問題12: Decision to renumber 1-4 → 65-68 relies on the section expectation; if a downstream consumer wants the literal printed local numbers, this would need revisiting.
• 問題13: Separator/bullet glyph (half-width . + space vs full-width ．) is inherently ambiguous in this low-contrast retypeset; chose half-width based on consistency with the half-width sentence terminator, but a full-width ． reading cannot be 100% excluded.
• 問題13: Whether a half-width space follows the 注意事項 bullet dot vs the cell separators is a fine formatting judgment; followed A (bullet: no space; separator: one space).
• 問題13: 「マスつき」(row ③), 「持参うえ」, 「9月9日」(row ④, vs the matching 9月19日 row) and the pasted クレジットカード instruction all appear to be source retype errors but were transcribed as printed per task rules.

Source-text quirks the OCR transcribed verbatim (printed typos/recall artifacts; the ones judged to be retype errors were corrected in the layer below) — 32 noted at OCR time:

• 問題1: Q6 (拒んで): the printed underline spans '上拒んで' (starting at 上 rather than at 拒); transcribed as printed. The intended target is almost certainly just 拒んで — likely a retype/underline-placement error in the source.
• 問題1: Underlined headwords transcribed as printed; no furigana present anywhere on the page.
• 問題5: The 問題5 instruction sentence was printed run-together with question 1 as list item '1.', i.e. '1.問題5　次の文の(　　)に入れるのに最もよいものを、1・2・3・4から一つ選びなさい。' immediately followed by 'この島では…'. The instruction was extracted into the instruction field; question 1's stem begins at 'この島では'.
• 問題5: Question 10 is a two-speaker dialogue (夫 / 妻); each speaker label is separated from its 「…」 line by a full-width space (夫　「…」), preserved as printed, one <p> per speaker.
• 問題5: Question 9 option 1 '悔やまないようにする' and question 10 option 1 '買ってきてはないんだ' are slightly odd wordings but transcribed exactly as printed (likely retype artifacts of the source document).
• 問題6: Q38 (=38) source wording reads 'からといって/かというと' duplication pattern; transcribed as printed.
• 問題6: Item 12: the sentence run reads 'それ＿＿＿＿＿_★__＿＿＿ない'; the ＿ slot run was placed after 'それ' and before 'ない' as printed.
• 問題7: Paragraph 5 prints a half-width exclamation mark before the closing quote: ではないか!」 (transcribed as printed; both transcribers had originally dropped it).
• 問題7: Option 44-3 reads 助かっていたかなんだ as printed; the wording is unusual but verified verbatim against the double-resolution scan.
• 問題7: Blank 44 is immediately followed by a 注3 reference: 牛のおかげで【44】、と私は気を良くし（注3）; transcribed as printed.
• 問題7: Inner-quote bracket usage varies in the source and is transcribed as printed: 「モーいいか」 (corner brackets) in paragraph 4, but 『モーいいか』 in paragraph 5; 『モーモー』 in paragraph 6.
• 問題8: Passage (2) is an embedded scan of a company website notice; its body text was transcribed from the scanned image (verified against the double-resolution p09-t/p09-b halves). The printed lead-in line above it reads '以下は、ある会社がホームページに掲載したお知らせである。' (school label '2.(2)'). The notice header reads 'ミツヌマ健康食品株式会社＞ニュース' with a '＞' separator; the company name is ミツヌマ (not ミツヌヌ).
• 問題8: Passage (3) retype artifacts, transcribed as printed: '執着を持っていることしたら' (likely intended としたら or こと（を）したら), and '諦めの中に懲ってしまう' (the printed glyph is 懲, where 凝ってしまう would be expected). The glyph is 懲, not the Chinese-simplified 懒.
• 問題9: Apparent retype oddities transcribed as printed: Q54 option 2 「現実には容易ではない」 reads slightly oddly after 「欠かせないが」; Q55 option 3 「経験を通して精巧になると考えられる」 reads oddly (expected 「習得する」 or similar); Q57 stem asks 「どのような時間か」 yet the options are sentence-form actions (e.g. 「働くことの意義を考える。」). None corrected.
• 問題9: Passage (3) is printed cleanly and is NOT corrupted (an earlier independent transcription introduced spurious typos here, e.g. 「決して「人生」である」, 「愚は1）に」, 「期間しか許された時間ではない」, 「どう答えるだろう？」; the page actually prints 「決して「人生」ではない」, 「仕事体験…といった愚（注1）にもつかない」, 「このわずかな期間にのみ許された、きわめて貴重な時間なのだ」, 「どう答えるだろう。」).
• 問題10: The passage uses 観賞 (not 鑑賞) consistently, and 出来上り (Q2 option 3) as printed; transcribed exactly as printed.
• 問題11: Passage A author/work line printed as 「（中村俊介『世界遺産が消えてゆく』による）」; passage B line 「（五十嵐太郎　『建築はいかに社会と回路をつなぐのか』による）」 has a full-width space after the author name (preserved as printed).
• 問題12: Q67 stem prints 'どうのような場合か' (a retype typo for 'どのような', confirmed against the double-resolution image); transcribed as printed.
• 問題12: Author in citation prints as 樋口祐一; the actual author is 樋口裕一, so 祐 is likely a retype error for 裕. Transcribed as printed.
• 問題12: Paragraph 5: 'それが書かれていないからと批判されても' on the image (A dropped 'から'; B had a corrupted tail).
• 問題12: Options are printed as unnumbered bullet lines (・); numbered 1-4 top-to-bottom as printed.
• 問題13: 問題13 instruction line was pasted from a different exam: it says 「クレジットカードの入会案内」, but the actual right-page document is a 就職支援行事（9月-12月）のスケジュール (job-hunting support events schedule). Transcribed the instruction as printed.
• 問題13: The instruction sentence ends with a half-width period "." instead of the usual full-width 。 — kept as printed.
• 問題13: The printed document numbers the two questions as plain unboxed 「1)」 and 「2)」 (this retypeset uses plain numbering); there are no boxed/printed numbers 69/70. Assigned numbers 69 and 70 per the section expectation.
• 問題13: Row ③ content reads 「広告. マスつき業界」; 「マスつき」 is almost certainly a retype typo for 「マスコミ」 (cf. the 合同企業説明会 sub-table which lists 「広告. マスコミ」). Kept as printed.
• 問題13: Throughout the 内容 column and the 合同企業説明会 sub-table, the separators within paired terms are printed as a half-width period plus space (e.g. 「金融. 証券」「貿易. 流通」「広告. マスコミ」「メーカー. IT」「教育. 福祉」), where a 中点 (・) or 読点 (、) would normally be expected. Verified against the double-resolution scan: the glyph is a small low half-width period, matching the half-width period used as the sentence terminator and the bullet glyph. Transcribed as printed.
• 問題13: Row ④ closing bracket is printed as a half-width ] (「基本的な情報]」) while its opening bracket and the rest of the document use full-width ［］. Verified on the scan; kept as printed.
• 問題13: The 注意事項 box uses a half-width period "." as the bullet glyph at the start of each item (not ・). Kept as printed.
• 問題13: Intro line reads 「学生証を持参うえ」, missing the expected の (持参のうえ); kept as printed.
• 問題13: Schedule grid has an empty leading header cell above the ①〜⑩ number column (header row: empty | 日 | 時間 | 行事名 | 内容); verified on the scan.
• 問題13: Row ④ prints 「10月16日（水）」 and 「内容は9月9日と同じです」 — note 9月9日 is itself likely a retype slip for 9月19日 (the matching 就職ガイダンス row), but kept exactly as printed.
• 問題13: Schedule title month range printed as 「9月-12月」 with a hyphen between the full-width 月 labels; transcribed as printed.

Corrections applied (15)

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

• 問1 q6 question: 「これ以<u>上拒んで</u>」 → 「これ以上<u>拒んで</u>」
  • 以上 is one word (いじょう); the reading target is just 拒んで (こばんで). The printed underline runs under 上拒ん, but the asked word is 拒んで.
  • evidence: underline-boundary-vs-image
• 問8 q48 passage: 「執着を持っていることしたら」 → 「執着を持っているとしたら」
  • 「執着を持っていることしたら」is ungrammatical; the correct connective is と (としたら, 'if/supposing'). Both witnesses independently read 「執着を持っているとしたら」.
  • evidence: studyinjapan docx witness reads '執着を持っているとしたら'; jlpt-freq witness reads '執着を持っているとしたら' [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q65 passage: 「（樋口祐一『差がつく読書』による）」 → 「（樋口裕一『差がつく読書』による）」
  • The author of 『差がつく読書』is 樋口裕一 (well-known 小論文 author); 祐 is a retype error for 裕.
  • evidence: Multiple bookstore listings (booklive, bookwalker, kinokuniya) and Wikipedia confirm 『差がつく読書』is by 樋口裕一. Both witnesses retain the 祐 typo, but the published-book metadata is decisive. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q67 question: 「本が良書と言えるのはどうのような場合か。」 → 「本が良書と言えるのはどのような場合か。」
  • 「どうのような」is a non-word; the interrogative is どのような. A retype inserted a stray う.
  • evidence: grammar: どうのような is ungrammatical; jlpt-freq witness reads '本が良書と言をるのはどのような場合か' [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「広告. マスつき業界」 → 「広告. マスコミ業界」
  • 「マスつき」is a non-word; the 合同企業説明会 sub-table in the same passage prints 「広告. マスコミ」and Q70 concerns マスコミ業界. Retype garbled マスコミ into マスつき.
  • evidence: internal consistency: the 合同企業説明会 sub-table prints '広告. マスコミ'; jlpt-freq witness Q70 reads '林さんはマスコミ関連の企業' [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「内容は9月9日と同じです」 → 「内容は9月19日と同じです」
  • Row ④ (就職ガイダンス) references the identical earlier 就職ガイダンス, which is row ① on 9月19日; no 9月9日 event exists. The parallel row ⑤ (留学生ガイダンス) references row ② (留学生ガイダンス, 9月24日), confirming the pattern. A retype dropped the 1 from 19.
  • evidence: internal consistency: row ① 就職ガイダンス is 9月19日; row ⑤ references '9月24日と同じ' matching row ② identically; no 9月9日 row exists [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「学生証を持参うえ、会場へ」 → 「学生証を持参のうえ、会場へ」
  • 「持参うえ」is missing the particle の; the pattern is 〜のうえ (持参のうえ = 'upon bringing'). A retype dropped の.
  • evidence: grammar: 〜のうえ requires の; 持参うえ is ungrammatical [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q48 passage: 「諦めの中に懲ってしまう」 → 「諦めの中に凝ってしまう」
  • 懲る(こりる/こらしめる) does not fit the sense 'to get fixated/stuck within resignation'; the intended verb is 凝る (こる) 'to become fixated'. The original Kushida Magoichi text reads 凝ってしまう. 懲→凝 is a homophone/glyph retype error already flagged in the quirks.
  • evidence: Category homophone, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q46 passage: 「大過(注)なく」 → 「大過（注）なく」
  • The inline note marker uses half-width parentheses while the gloss line in the same passage (（注）大過：…) and all other note markers in the exam use full-width （）. The half-width form is a retype/OCR artifact; the JLPT exam uses full-width consistently.
  • evidence: Category orthography, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問10 q59 passage: 「のだ、(中略)それを大事そうに」 → 「のだ、（中略）それを大事そうに」
  • This 中略 marker is half-width while the four other 中略 markers in the exam are full-width （中略）. Half-width is a retype artifact.
  • evidence: Category orthography, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問10 q59 passage: 「習練(注1)」 → 「習練（注1）」
  • Inline note marker uses half-width parentheses; the gloss line （注1）習練：練習 and exam convention use full-width. Retype artifact.
  • evidence: Category orthography, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問10 q59 passage: 「タッチ(注2)」 → 「タッチ（注2）」
  • Inline note marker uses half-width parentheses; the gloss line （注2）タッチ：… and exam convention use full-width. Retype artifact.
  • evidence: Category orthography, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問10 q59 passage: 「大向こう(注3)」 → 「大向こう（注3）」
  • Inline note marker uses half-width parentheses; the gloss line （注3）大向こう：… and exam convention use full-width. Retype artifact.
  • evidence: Category orthography, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問12 q65 passage: 「酷評する(注)書評」 → 「酷評する（注）書評」
  • Inline note marker uses half-width parentheses; the gloss line （注）酷評する：… and exam convention use full-width. Retype artifact.
  • evidence: Category orthography, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q0 passage: 「（注1）にもつかない教」 → 「にもつかない（注1）教」
  • （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]