N1 2014-07 — 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: Cleanly retypeset digital text (Word-style), crisp and fully legible, but very dense layout with small type since original booklet pages are compressed; Chinese watermark headers ('小程序：日语轻松考', '祝亲考级 通过~通过~通过~') sit in page margins without overlapping content; grayscale only.

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. 88 leaf-level differences were examined in total.

Residual OCR risks (27):

• 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: Source is a third-party Chinese retypeset; dense digital text was crisp so confidence is high, but any retype-introduced typo would be faithfully reproduced and cannot be distinguished from the original here.
• 問題2: Fill-in blanks normalized to （　　） per convention regardless of the printed blank widths, which vary on the page.
• 問題3: No image-processing tools were available to magnify regions; adjudication relied on the high-resolution half-page (p02-b/p03-t) and full-page (p02) renders, both of which were legible for every contested character.
• 問題3: Q16 ending です vs のだ: chose です as printed; the underline boundary (ストレート only) is the more consequential call and is unambiguous in the image.
• 問題4: None significant. All 6 questions and 24 options visually confirmed against the double-resolution renders; underlined target words match printed underlines throughout. No furigana, tables, or worked examples present in this section.
• 問題5: Q27 option 2 浴びたにもかかわらず: A flagged a stray superscript mark in its full-page render; the p04-t render shows clean text, so this is judged a print/render artifact and omitted (both transcribers agree on the text).
• 問題5: All differences resolved to B, which worked from the double-resolution half-page renders; the same crisp renders were used for adjudication, so a shared rendering artifact (rather than a true source feature) cannot be fully excluded, though none was observed.
• 問題6: The leading 'X' before Q40 is almost certainly a retypeset artifact/placeholder rather than exam text, but it is printed and was kept verbatim per the transcribe-as-printed rule; a downstream consumer may want to strip it.
• 問題6: The instruction-line slot run is rendered with the conventional ' _ _ ★ _ ' four-slot pattern; the printed run is a generic underline-with-★ whose exact slot count is stylized, so this is a convention-driven normalization rather than a literal slot count.
• 問題7: The 「2008 年」 space may be a retypeset spacing artifact, but it is clearly visible in the half-page render and is transcribed as printed.
• 問題8: Passage (3) attribution 「著者代表森亘」 is an unusual/likely-corrupted retype of the real source attribution but is transcribed exactly as printed; reading is confident at this resolution.
• 問題8: No <u> underlines appear in passages (2)/(3) where standard JLPT printings underline the referenced phrases; retypeset omits them, transcribed without <u> as printed.
• 問題8: p07-b is 問題9, not part of 問題8; confirmed 問題8 spans only p06 and the top of p07. Section boundary is clean.
• 問題9: 本来 vs 未来: the underlined glyph in passage (1) para 4 and Q51 stem is read as 未来 from p07-b, but 本/未 are close at this print size; 本来 would be the semantically expected word, so a sharper scan could overturn this.
• 問題9: Q50 stem と: the standard 'とあるが' frame is retained, but the と is faint in print and A flagged it as possibly absent ('表れあるが'). Both transcribers wrote とあるが, so it is kept, but a residual chance the retype literally omits と.
• 問題9: Q51 stem どのような vs どういう: read as どのような (A) from p07-b kana spacing; borderline at this resolution.
• 問題10: Paragraph 4 reads '豊かだったそうではなかったりする自然がある' which is grammatically odd and may be a retypeset corruption of '豊かだったりそうではなかったりする'; transcribed as printed per convention.
• 問題10: The retypeset has no furigana and a very dense layout; remaining single-kana misreads in long prose lines cannot be fully excluded, though disputed loci were all re-read at double resolution.
• 問題11: Digits 15 and 192 are rendered half-width ASCII per convention; the retypeset prints them full-width — intentional normalization, not an error.
• 問題11: The '選択肢' wording in Q64 option 1 is semantically odd but is faithfully transcribed as printed in this third-party retypeset.
• 問題11: Passage A title uses 子供 while its body uses 子ども; kept as printed (inconsistency is in the source).
• 問題12: Paragraph 1's exact wording (especially 'あたかも必要不可欠とばかり付加して製品を売り込もうとし' and 'かのごとく錯覚') was read directly from the double-resolution scan; the retypeset is dense, so a residual small-kana misread is conceivable but the high-zoom crops were clear.
• 問題12: The 終行 in （注1）is transcribed as printed (likely a retype typo for 修行); deliberately not corrected per conventions.
• 問題13: Rendering the rowspan-style 年会費/旅行保険 cells without rowspan attributes yields ragged rows (5 cells vs 6); downstream consumers expecting uniform column counts may need to account for the spanning header.
• 問題13: The ID-document line spacing (single vs double full-width space) is a judgment call about tabular gaps; chose single for consistency, but the print shows wide gaps.
• 問題13: Several values are retyped typos (最高2.000万円 period, 最高 3,000万円 internal space, 誘えない); transcribed as printed per convention but not original to the official exam.

Source-text quirks the OCR transcribed verbatim (printed typos/recall artifacts; the ones judged to be retype errors were corrected in the layer below) — 21 noted at OCR time:

• 問題2: 問題2 instruction is printed without a trailing 。 (sentence period), unlike the other 問題 instructions; transcribed as printed.
• 問題4: 問題4 questions 20-25 (6 questions), 語彙の使い方 section. Source is the 日语轻松考 retypeset; transcribed as printed.
• 問題4: Q23 option 3 ends with 「元気が出ました」 (verified against both half-page renders); A's 「元気が出てきた」 was a misread.
• 問題5: 問題5 instruction is printed as 「入れるのは最もよいものを」 (typical wording is 「入れるのに」); transcribed as printed.
• 問題6: Q40 begins with a printed 'X ' before 交通のバス路線 (between the boxed number [40] and the sentence). Both transcribers and the double-resolution images confirm it; transcribed as printed. Likely a stray retypeset artifact / placeholder rather than part of the sentence.
• 問題6: Instruction line uses the 問題6 slot pattern (_ _ ★ _) as printed, not a bare ★.
• 問題7: Source citation printed as 「2008 年2月号」 with a space between 2008 and 年; retained as printed.
• 問題8: Passage (1) email body prints 「大変込み合います」 (込, not 混) and 「特に込み合う」 in Q46 option 4; transcribed as printed. Phone number printed with full-width hyphens (031－234－5544) normalized to ASCII '-' and full-width digits to half-width per conventions; wave dash 10〜17時 is U+301C.
• 問題8: Question stems 47 (「ああっ！」と叫んだ) and 48 (思い悩まずにいられる) reference phrases that in standard JLPT printings are underlined within the passage body, but this retypeset prints no underline in either the passage or the stem; transcribed without <u> as printed.
• 問題8: Passage (3) source citation prints the author as 著者代表森亘 (森亘); this is an unusual attribution, transcribed exactly as printed.
• 問題9: Passage (1) and Q51 stem: the underlined reference reads ②未来の機能 in both the passage body and the question stem (the glyph is 未, not 本; verified on the double-res render). Semantically odd, but transcribed as printed in this retypeset.
• 問題9: Q54 stem prints 'ノウハウ本あるもの' whereas the passage (2) body prints 'ノウハウ本なるもの' (あるもの vs なるもの); both transcribed as printed — likely a source typo in the stem, which carries no underline.
• 問題10: Passage paragraph 4 contains "豊かだったそうではなかったりする自然がある", which reads oddly (possibly a retypeset corruption of something like "豊かだったりそうではなかったりする自然がある"); transcribed as printed.
• 問題10: Q60 option 3 uses kana "ひかれる" (not 引かれる/惹かれる) as printed.
• 問題11: Q64 option 1 reads '選択肢に含めたほうがよい' as printed; '選択肢' is an odd word choice in context (one might expect '蔵書' or similar), but it is transcribed exactly as the retypeset prints it.
• 問題11: Passage A author is printed as 若葉千佳子 with the title 『子供の本棚』 (using 子供, while the body text uses 子ども); transcribed as printed.
• 問題12: ②不可能である is underlined in full (not just 不可能); the ② circle is outside the underline. Verified in both the passage (p11-b) and the question 66 stem (p12-t).
• 問題12: （注1）修道僧：終行中の僧 — '終行' is likely a retype typo for '修行', but transcribed as printed.
• 問題12: Source citation prints as （池内了『生きのびるための科学』による）; transcribed as printed.
• 問題13: Suspected source typos transcribed as printed: D-column 旅行保険 values '最高2.000万円' use a full stop instead of a comma (both 国内 and 海外 rows). C-column 旅行保険 海外 is printed with an internal space: '最高 3,000万円'.
• 問題13: Suspected typo in 'お申込み方法' step 3: '当社の判断で、ご希望に誘えない場合もございます。' — '誘えない' appears to be an error for '添えない'/'沿えない', transcribed as printed.

Corrections applied (13)

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

• 問3 q14 question: 「<u>無償</u>で受けられます」 → 「<u>無償で</u>受けられます」
  • The correct choice ただで is a complete adverbial. Leaving で outside the tested span produces ただでで受けられます; the full target is 無償で.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問3 q16 question: 「<u>ストレート</u>に言って」 → 「<u>ストレートに</u>言って」
  • All four choices already end in に, so leaving the source retype's に outside the tested span duplicates the particle. The full target is ストレートに.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問3 q17 question: 「<u>お手上げ</u>だ」 → 「<u>お手上げだ</u>」
  • The correct choice どうしようもない is a complete predicate; leaving だ outside the tested span produces どうしようもないだ. The full target is お手上げだ.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問3 q18 question: 「<u>格段</u>に増えている」 → 「<u>格段に</u>増えている」
  • All four choices already end in に, so leaving the source retype's に outside the tested span duplicates the particle. The full target is 格段に.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問6 q40 question: 「X 交通のバス路線が一部廃止される問題で」 → 「交通のバス路線が一部廃止される問題で」
  • The leading 'X ' is a retypeset placeholder artifact, not exam text. Quirks note records that it is printed between the boxed [40] and the sentence and is almost certainly a stray placeholder; the grammatical sentence begins at 交通. No in-sentence company name is used (the firm is referred to as 会社側).
  • evidence: quirks note explicitly flags it as a retypeset artifact a downstream consumer may strip; both OCR witnesses (mondai6.A/B) and the freq witness carry the same 'X ' from the shared source family. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q54 question: 「ノウハウ本あるもの」 → 「ノウハウ本なるもの」
  • The stem quotes the passage, whose body reads 'ノウハウ本なるものをまったく信用していない'. The stem's 'あるもの' is a retype error for 'なるもの'; JLPT stems reproduce the passage phrase exactly.
  • evidence: passage body in this same JSON reads 'ノウハウ本なるものをまったく信用していない' (柳井正『現実を視よ』). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q59 passage: 「豊かだったそうではなかったりする自然がある」 → 「豊かだったりそうではなかったりする自然がある」
  • The parallel 〜たり〜たりする construction requires たり on the first clause too. The current text drops the first り, leaving the ungrammatical '豊かだった' + 'そうではなかったりする'. The author uses 〜たり〜たり repeatedly in this passage (緑だったり茶色だったり, 見えたり…迫ってきたり).
  • evidence: Google Books (角田光代) snippet confirms 'そうではなかったりする自然がある'; grammar requires たり on both enumerated clauses. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q65 passage: 「終行中の僧」 → 「修行中の僧」
  • '終行' is not a word. The gloss defines 修道僧 (a monk), who engages in 修行; the natural gloss is '修行中の僧'. 終→修 is a confusable retype error.
  • evidence: gloss text reads '（注1）修道僧：終行中の僧'; 修道僧 is glossed as a monk; 終行 is a clear non-word. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「ご希望に誘えない」 → 「ご希望に添えない」
  • 'ご希望に誘えない' is semantically impossible (誘える = to invite). The fixed formal phrase is 'ご希望に添えない/沿えない場合もございます'. 誘→添 is a retype error of the set idiom.
  • evidence: standard business idiom is ご希望に添えない; 誘えない does not parse here. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「最高3,000万円</td><td>最高2.000万円」 → 「最高3,000万円</td><td>最高2,000万円」
  • 国内 row D-column uses a full stop '2.000万円' where every other monetary cell uses a comma thousands separator (5,000/3,000/8,000万円). The '.' is a retype/OCR typo for ','.
  • evidence: internal consistency: all sibling cells use comma separators; Japanese amounts never use a period as a thousands separator. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「最高 3,000万円</td><td>最高2.000万円」 → 「最高3,000万円</td><td>最高2,000万円」
  • 海外 row: C-column has a stray space ('最高 3,000万円') and D-column uses a full stop ('2.000万円'). Sibling cells have no internal space and use comma separators; both are retype artifacts.
  • evidence: internal consistency: sibling cells read 最高300万円/最高5,000万円 with no internal space and comma separators. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q51 passage: 「適合マークが②<u>未来の機能</u>を果たすには」 → 「適合マークが②<u>本来の機能</u>を果たすには」
  • 「本来の機能を果たす」is a fixed collocation; 「未来の機能を果たす」is not idiomatic and semantically incoherent here (the passage discusses the mark fulfilling its intended/proper function). 未/本 are a classic OCR confusion at small print sizes, and the transcriber explicitly flagged this glyph as ambiguous (quirk lines 16/19). Q51's correct answer describes the function the mark properly serves.
  • evidence: Category homophone, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q51 question: 「②<u>未来の機能</u>とあるが」 → 「②<u>本来の機能</u>とあるが」
  • Same underlined target phrase as in the passage body; must read identically. 本来の機能 is the idiomatic, semantically correct form (本/未 OCR confusion, flagged by transcriber as ambiguous).
  • evidence: Category homophone, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]