N1 2017-07 — 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: Mostly crisp retypeset digital text with no furigana issues; watermarks stay in page margins. Docked one point because the 問題13 info panel (pages 12-13) is a low-resolution Word screenshot with stray formatting marks and an inset box that occludes a few underlying lines.

Every section was transcribed twice independently (a full-page pass and a double-resolution half-page pass) and reconciled. Of 13 sections, 2 agreed on all content between the two passes (reconciled automatically, differing only in transcriber notes) and 11 had at least one content difference resolved by a third adjudication pass that re-read the page images. 91 leaf-level differences were examined in total.

Residual OCR risks (27):

• 問題1: Section contains only 問題1 (questions 1-6), matching expectations; no structural ambiguity found.
• 問題1: All readings are kana-only and the render is crisp digital text, so dakuten/handakuten and long-vowel marks were the only real risk; each was checked directly.
• 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: All six questions are simple single-underline vocabulary items with no furigana, tables, or passages; risk is low. No illegible spots in the mondai3 region of the renders.
• 問題4: The 発足 underline in Q23 option 4 is faint at this resolution; restored on the basis of the visible underline plus the consistent convention that the headword is underlined in every option. Confidence high but not absolute.
• 問題4: No 問題13 information panel or worked example appears in this section (mondai4 is vocabulary-usage 20-25 only), so the low-res screenshot caveat does not apply here.
• 問題5: Q34 option 2 "はすだ" is preserved as a printed typo (expected はずだ); transcribed verbatim per conventions, but it is an intentional-looking source typo rather than an OCR artifact.
• 問題5: Q33 stem particle に (に失敗だった) and the 新入生…失敗 phrasing read oddly but are transcribed exactly as printed in this third-party retypeset.
• 問題6: Q38 '彼女以上に優れている' has no clear antecedent in the stem; transcribed as printed per the no-correction rule but could be a source retype error.
• 問題6: The exact number of low-line glyphs in the instruction's '★' run is hard to count at this resolution; rendered as three full-width low lines per side following the ＿＿＿ convention. The semantic content (instruction text) is unaffected.
• 問題7: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題8: The 問題8 instruction line is reconstructed, not transcribed (it lives on p05, outside the supplied images). High confidence on wording from the parallel 問題9 line, but it should be verified against p05 if/when available.
• 問題8: Passage (1) lead-in '以下は…' is treated as the first <p> because it sits inside the box; if the project intends lead-ins inside a boxed document to be styled differently, revisit — but this matches the stated convention.
• 問題9: Passages (2) and (3) are set in very small dense type; A's readings match every legible token but a few low-frequency kanji (e.g. 辿/緩) rely on context plus partial glyph shape.
• 問題9: Passage (1) para 3 「よりの」 is kept as printed; it is almost certainly a retype typo for 「よいの」 but the glyph in p07-b reads より.
• 問題9: Passage (2) citation dash transcribed as a single U+2015 (―) vs passage (1)/(3) double U+2015; the single-dash judgment is based on the short mark length in p08-t and could in principle be a U+2014/U+FF0D in the source.
• 問題9: The exact left/right extent of the underline under Q54 「小さく説く」 (inner phrase vs including quote marks) was taken to match the body usage (inner phrase only).
• 問題10: 瑕跡 (kizu-ato, 'flaw mark') in 'その仕上がりは瑕跡ではなく' is read from a 3x crop and is plausible but the 瑕 character is dense; the surrounding phrase '景色に見立てることができるほど美しい' is clear, so the reading is contextually sound.
• 問題10: 杯 in '義父が好んで使っていたという杯' is a single small glyph at end of a clause; read as 杯 (sakazuki) from the magnified crop and contextually fits (a small drinking vessel among 小鉢), but is the least-certain single character.
• 問題10: The exact underline extents of ② and ③ in the first half were inferred from the question stems plus the crisp ① crop; ② and ③ underlines themselves sit in the denser band, but their text matches the verbatim question stems on p10-t.
• 問題11: Passage B is a reconstructed hybrid; while every clause was checked on the double-res crop, the retypeset source itself may contain typos relative to the official exam, and a few particle-level choices (は vs ほど, 上で) rely on single-glyph reads.
• 問題11: The p11-t half-page render is the highest resolution available; no independent third image exists to triangulate, so passage B rests on a single (clear) source read.
• 問題12: 履し (Q67 opt 2) is a retype typo for 覆し; preserved as printed per convention — downstream consumers should be aware it is intentional.
• 問題12: The 'しかし、と' / 'きに科学は' paragraph split is preserved as two <p> because the source prints a paragraph indent there, but it is a retypeset artifact splitting the word とき; the original is a single paragraph.
• 問題12: The passage is dense; while every span was zoom-verified, a handful of common kanji in a 4-paragraph block carry the usual small residual OCR risk.
• 問題13: The 料金表 ご入会条件 column for 平日A/B/C is empty in the source; the screenshot's cell borders make it ambiguous whether 16歳以上の方 is meant to span those rows. Transcribed as empty <td> per the literal layout.
• 問題13: Whether the full-width time colon ：should be normalized like ASCII digits is a borderline call under the conventions; it was kept full-width because the convention lists colons under Japanese punctuation behavior and the glyph is unambiguously full-width on the page.

Source-text quirks the OCR transcribed verbatim (printed typos/recall artifacts; the ones judged to be retype errors were corrected in the layer below) — 18 noted at OCR time:

• 問題4: Q22 滅びる is the printed headword (horobiru); options use 滅びる/滅びて/滅びた forms. Option 1 reads 資金が滅びる (the natural verb would be 尽きる/不足する); transcribed as printed (this is a distractor).
• 問題4: Q25 option 2 underline spans 見落と only (見落とされて with the passive ending されて outside the underline), as printed.
• 問題5: Q33 stem reads "昨日の新入生歓迎会に失敗だった" (に instead of the expected は), which is grammatically odd; transcribed as printed.
• 問題5: Q34 option 2 is printed "待つこともあるはすだ" — "はすだ" appears to be a typo for "はずだ"; transcribed as printed.
• 問題6: Q38: option 1 '彼女以上に優れている' is printed as-is; '彼女' has no antecedent in the stem and may be a retype oddity, but is transcribed as printed.
• 問題8: Passage (1) heading uses a half-width '>' as the separator: 関川薬品工業株式会社>お知らせ — transcribed as printed.
• 問題10: Citation prints the year-range as 2014年10-11月号; digits and the hyphen normalized to half-width ASCII. 暮しの手帖 printed without okurigana 'らし', kept as printed.
• 問題10: Punctuation inside the お料理屋さん quote kept as printed: full-width 。 before the closing 」.
• 問題11: Passage A citation reads 「（文化庁2015年10月23日取得による）」 as printed; '取得' is unusual for a citation but transcribed as printed. Verified char-by-char against p10-b.
• 問題12: The 問題12 passage spans the bottom of p11 into the top of p12. Verified character-by-character against the double-resolution half-page renders (p11-b, p12-t, p12-b). Transcriber A's passage was heavily garbled/ungrammatical (likely a misread of the dense low-resolution full-page render); transcriber B's reading matches the image exactly and was adopted.
• 問題12: Q67 option 2: the source prints '見方を履し変化をもたらした'. 履し is almost certainly a retype error for '覆し' (the glyph clearly shows the 尸 radical of 履, not the 襾 top of 覆), but it is transcribed as printed per convention.
• 問題12: Passage: the source breaks the word 'とき' across a paragraph boundary — a printed line ends with '…しかし、と' and the next printed line begins with a full-width paragraph indent before 'きに科学は…'. Because the indent marks a new printed paragraph, it is rendered as two <p>; this is almost certainly a retypeset artifact (the original reads '…しかし、ときに科学は…' as one paragraph).
• 問題13: The printed 問題13 instruction sentence reads 「学生レポーター」の募集案内 (a student-reporter recruitment notice on a city-hall homepage), but the actual document is a りんどうスポーツジム sports-gym membership announcement (ご入会案内 / 夏のご入会キャンペーン). This is a mismatch in the source retypeset; the instruction is transcribed as printed.
• 問題13: Time separators are printed as full-width colons (e.g. 8：00〜23：00); transcribed as printed.
• 問題13: 「キャッシャカード」 in 入会手続き ② is printed as-is (suspected typo for キャッシュカード); transcribed as printed.
• 問題13: 「お越してください」 in 入会手続き is printed as お越して (would normally be お越しください); transcribed as printed.
• 問題13: 「お問い合わせてください」 in the キャンペーン panel is printed as お問い合わせて (would normally be お問い合わせください); transcribed as printed.
• 問題13: In the 料金表 table, only フルタイム (16歳以上の方) and シニア (60歳以上の方) show a ご入会条件 value; the 平日A/平日B/平日C rows' ご入会条件 cells are left empty as printed (the column border visually merges these rows in the screenshot).

Corrections applied (20)

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

• 問3 q18 question: 「<u>むっとした</u>ようだった」 → 「<u>むっとしたようだった</u>」
  • The printed underline runs continuously under the entire phrase むっとしたようだった; our span underlines only むっとした. The answer choices are full phrases (驚いたような顔をしていた etc.), confirming the whole phrase is the target.
  • evidence: underline-boundary-vs-image
• 問4 q25 answers[1]: 「<u>見落と</u>されて」 → 「<u>見落とされて</u>」
  • The printed underline covers 見落とされて; our span drops されて. Headword is 見落とす; this option is the passive 見落とされて, and the print underlines the full conjugated form.
  • evidence: underline-boundary-vs-image
• 問1 q1 answers[3]: 「いや」 → 「いやす」
  • Reading-of-verb question for 潤す(うるおす); the other three answers are す-ending verbs (たがやす/耕す, ぬらす/濡らす, うるおす/潤す). 'いや' is a truncated retype of 'いやす'(癒す) and is not a verb, breaking the answer pattern.
  • evidence: All sibling distractors are す-ending verbs; 'いや' alone is not a valid verb reading. (jlpt-freq witness for this exam is the garbled listening section and does not corroborate, but the intrinsic answer-pattern argument is conclusive.) [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q67 answers[1]: 「見方を履し変化」 → 「見方を覆し変化」
  • '履し' is a non-word; the intended verb is 覆し(くつがえし, 'overturn'). This is the correct answer (correctAnswerIndex 1), so it must be valid Japanese. 覆 was misread as the visually dense 履.
  • evidence: Google Books HIT for 'ものの見方を覆して' (國文學); 'ものの見方を履して' returns no hits. Passage context 一変させ supports 'overturned'. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問5 q34 answers[1]: 「待つこともあるはすだ」 → 「待つこともあるはずだ」
  • 'はすだ' is a non-word dakuten-drop typo (ず→す); the grammatical form is はずだ. As a mondai5 cloze distractor it must be valid Japanese, contextually wrong by meaning, not broken grammar.
  • evidence: Clear non-word; はず is the only valid form. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q70 passage: 「キャッシャカード」 → 「キャッシュカード」
  • 'キャッシャカード' is a non-word; intended term is キャッシュカード (cash card). Classic ャ/ュ kana-swap transcription artifact.
  • evidence: Clear non-word; キャッシュカード is the standard term. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q69 passage: 「受付にお越してください」 → 「受付にお越しください」
  • Honorific request set phrase is お+masu-stem+ください; the inserted て (お越してください) is ungrammatical. JLPT 問題13 notices are professionally edited and test content retrieval, not honorific errors, so this is a transcription artifact.
  • evidence: Google Books HIT for '受付にお越しください'; '受付にお越してください' returns no hits. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q70 passage: 「お問い合わせてください」 → 「お問い合わせください」
  • Same ungrammatical て-insertion in an honorific set phrase; correct form is お問い合わせください. Repeated identical artifact alongside お越して indicates transcription drift.
  • evidence: Honorific お+masu-stem+ください pattern; the て form is ungrammatical and unattested. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q50 passage: 「どのような自分を語り出していけばよりのかがわからない」 → 「どのような自分を語り出していけばよいのかがわからない」
  • 「よりの」is a non-word here; the grammatical form required is 「よいのか」(よい + の + か). The verb in the passage is 語り出していけばよい (it is good to ...). 「より」(more/from) makes no sense after the conditional 〜ば and before のか. Classic retype typo い→り.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q50 passage: 「（榎木博明『（ほんとうの自分）のつくり方――自己物語の心理学』による）」 → 「（榎本博明『（ほんとうの自分）のつくり方――自己物語の心理学』による）」
  • The author of 『〈ほんとうの自分〉のつくり方――自己物語の心理学』 is the psychologist 榎本博明 (Enomoto Hiroaki). 「榎木」is a retype error for 「榎本」(本→木 is a common visual/retype slip). Wrong real-world referent.
  • evidence: Category referent, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問12 q65 passage: 「金を通しての科学者と社会の網引き（注6）状態と言えるだろうか」 → 「金を通しての科学者と社会の綱引き（注6）状態と言えるだろうか」
  • The (注6) gloss defines the word as 「引っ張り合い状態」(a tug-of-war / pulling against each other), which is 綱引き (つなひき, rope-pulling). 「網引き」(net-pulling) is the wrong kanji; 綱 (rope) was retyped as 網 (net), visually near-identical kanji. The metaphor of science vs society pulling over money is unambiguously 綱引き.
  • evidence: Category homophone, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問12 q65 passage: 「（注6）網引き状態：ここでは、引っ張り合い状態」 → 「（注6）綱引き状態：ここでは、引っ張り合い状態」
  • Same error in the gloss line itself: the headword being glossed as 引っ張り合い (tug-of-war) is 綱引き, not 網引き. 綱→網 retype error.
  • evidence: Category homophone, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q0 passage: 「（注2）的なエ」 → 「的な（注2）エ」
  • （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9 q0 passage: 「（注3）にし」 → 「に（注3）し」
  • （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9 q0 passage: 「（注2）で、」 → 「で（注2）、」
  • （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9 q0 passage: 「（注3）に暮れるよ」 → 「に暮れる（注3）よ」
  • （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12 q0 passage: 「（注3）するか」 → 「する（注3）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]
• 問12 q0 passage: 「（注6）状態と」 → 「状態（注6）と」
  • （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 correctOrder: 「null」 → 「[1,3,2,0]」
  • The accepted sentence is ペンギンの赤ちゃんが生まれたこともあって赤ちゃんを一目見ようという多くの人で..., i.e. order 2431. With the stored ★ in the third slot, this confirms answer 3 and resolves the stale solver-order conflict.
  • evidence: Huanxing 2017-07 N1 grammar analysis gives q36 正解:3 and order 2431; see https://m.huanxingedu.com/news_item-72580.html. [order-conflict cleanup, 2026-06-24]
• 問6 q40 correctAnswerIndex/correctOrder: 「answer 1; correctOrder null」 → 「answer 4; correctOrder [2,0,3,1]」
  • The natural and externally attested assembly is デザイン画を描いては直すを繰り返すこと数十回, so ★ is option 4. The answer-key-only sources voting 1 are treated as row/key mistakes for this item.
  • evidence: Yahoo Chiebukuro discussion quotes the full sentence as デザイン画を描いては直すを繰り返すこと数十回, and the best answer explains the quoted phrase as valid Japanese; see https://detail.chiebukuro.yahoo.co.jp/qa/question_detail/q12232537222. [order-conflict cleanup, 2026-06-24]