N1 2021-12 — audit report

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

OCR confidence

Source: recall-retype — quality 3.5/5: Clean digitally typeset text (no scan artifacts), so individual characters are highly legible throughout; written half is sans-serif, listening half a separate serif Word-style document. Marked down for recall-fidelity problems: scattered typos and garbled phrases in stems and passages, no furigana, missing instructions/example blocks, one cloze blank and apparently two listening items dropped, and the gray diagonal 'Tôi Yêu Ngoại Ngữ Group / Yuuki Bùi' watermark on every page (it collides with the page-1 title and fills one listening answer-grid cell, though body text stays readable).

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. 154 leaf-level differences were examined in total.

Residual OCR risks (33):

• 問題1: The exact intended glyph of the q4 option-4 mark cannot be determined from a recall retypeset; preserved as ASCII apostrophe per the transcribe-exactly rule, but its origin (typo vs OCR/typeset artifact in the source PDF) is unknowable.
• 問題1: No external answer key consulted (out of scope); all correctAnswerIndex remain null as required.
• 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: The underline endpoints are derived from a recall-reconstruction retypeset; if the original exam underlined only the headword and the retypeset widened the rule, the printed source still shows the full group underlined, which is what was transcribed per the transcribe-exactly convention.
• 問題4: q22 option 3 「（押収）しまうので」 is grammatically odd (likely missing 「して」); transcribed as printed, but it may be a recall corruption of the original sentence.
• 問題4: q25 headword/option mismatch (必 vs 秘, and dictionary-form 秘める inside example sentences) is almost certainly recall/retype corruption; transcribed as printed and flagged.
• 問題5: The half-width ? characters were normalized to full-width ？ on the reading that ？ counts as Japanese punctuation under the convention; if the project intends to preserve printed half-width ? literally, q30 would need ? instead.
• 問題5: q30 option1 言えたけど and stem 言ったつけ are likely recall-version typos; transcribed as printed per instructions, not corrected.
• 問題5: q28 あっか (悪化 in kana) is a probable recall artifact but is what is printed; both transcribers agree.
• 問題5: Inner-word spaces were uniformly treated as justification artifacts. This is consistent across q26/27/28/31/33/36, but if any one of those were an intentional printed space the removal would be wrong; visual evidence (e.g. one-word 登場する being split) strongly supports the artifact reading.
• 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 後 しして gap in paragraph 5 is rendered as exactly one half-width space; the original dropped characters (likely 後回しにして) cannot be recovered and the precise printed gap width is approximate.
• 問題7: Source watermark text (Tôi Yêu Ngoại Ngữ Group / Yuuki Bùi) overlaps the passage diagonally; faint strokes could in principle hide a stray glyph, though column inspection found none.
• 問題7: This is a community recall reconstruction (回忆版), so several legible-but-likely-wrong tokens (付加, 地位置, the duplicated q43 option 2) reflect source errors, not OCR errors; they are transcribed verbatim per convention and may not match the genuine official exam text.
• 問題8: The 注-line colons are physically narrow/half-width glyphs followed by a space; both transcribers normalized them to full-width ： per the gloss convention. If the convention intended preserving a literal half-width colon, these would need revisiting (low risk — convention explicitly specifies the ：format for 注 glosses).
• 問題8: The stray tick after など in passage (4)'s 注 line is ambiguous (apostrophe vs. stray ＇/comma/printing artifact); rendered as ASCII apostrophe per both transcribers. Its exact intended character is unknowable from a recall retype.
• 問題8: Multiple in-passage/option typos (ブラス, 美意義, 工具も持つ, なぜ国によって, 起業社, 人並びへた, ビころか, 限りないく, 合理的より/合理的の) are preserved verbatim per the transcribe-as-printed rule; all confirmed against the double-resolution images, but they remain recall-version errors rather than authoritative exam text.
• 問題9: Passage（3）first sentence '販売されて略る': the verb ending 略る is a garbled glyph cluster (likely いる/おる) printed under the watermark; transcribed as the closest read but the exact characters are uncertain.
• 問題9: Passage（3）'朗読CDは趣味や実用に': 趣味 is the best read and is supported by p10, but the watermark crosses these kanji so a small residual risk remains.
• 問題9: Passage（2）'柔軟いにし' and '与える。'(maru): visually clear at double resolution, but these are odd enough that they could be source-OCR artifacts rather than intentional print; retained verbatim per the transcribe-exactly rule.
• 問題10: The injected Latin cluster ĨIU: the first glyph is a cap-height I with a tilde, read as Ĩ (U+0128). The tilde could in principle belong to a different code point, but Ĩ is the closest visual match; remaining uncertainty is in the exact Unicode identity, not the rendered shape.
• 問題10: 才艮本 first glyph: kanji 才 vs katakana オ are visually indistinguishable in this retype font; 才 chosen by both transcribers and retained.
• 問題10: Author name in the citation is genuinely mangled in the source (only 「二 郎」 with an internal space survives); transcribed as printed, true name unrecoverable from this image.
• 問題11: Passage B 「弊店の客層に合うよう 階または2階の」: a digit-sized gap before 階 with no visible glyph even at high zoom; transcribed as a single half-width space (presumed lost 1). If the source intended 「1階」 the space is a faithful representation of the printed gap, not the intended digit.
• 問題11: Passage B 「直9相談」: the glyph after 直 is read as the digit 9 by both transcribers and looks like 9 in the image, but it is a garbled retype (likely 直接ご); kept as 9 per transcribe-as-printed.
• 問題11: The watermark (Tôi Yêu Ngoại Ngữ / Yuuki Bùi) overlaps passage text diagonally; readings were taken from the clearer double-resolution halves, but a few watermark-crossed characters carry slightly lower confidence.
• 問題12: Passage token まさし: the double-resolution crop shows ま+さ(no dakuten)+し with no な; read as まさし, but the diagonal watermark grazes this run so a hidden dakuten on さ cannot be 100% excluded.
• 問題12: Dakuten on 「生まれるのてはなく」: て appears without dakuten versus surrounding が, but at body-text size sub-pixel dakuten loss is possible; kept as printed (て) per the transcribe-exactly rule.
• 問題12: Source is a community recall reconstruction (回忆版): the passage itself contains genuine retype/recall corruptions (方型, double-が sentence, ひつよう in kana) that are faithfully preserved, not corrected.
• 問題13: Exact width of the breadcrumb '>' separators and the spaces inside ［各教室 1 台］ is somewhat ambiguous in the handwriting-style retype font; transcribed as half-width > and with the visually-present spaces respectively.
• 問題13: The retype's ASCII spacing is intrinsically inconsistent (row ① unspaced vs rows ②–④ spaced; 7 日前 vs 3日前); preserved exactly as printed, but a downstream consumer relying on uniform spacing should be aware.
• 問題13: The projector header <br> reflects a cell line-wrap rather than a semantically forced break; kept to mirror the printed two-line layout but it is presentational.

Source-text quirks the OCR transcribed verbatim (printed typos/recall artifacts; the ones judged to be retype errors were corrected in the layer below) — 82 noted at OCR time:

• 問題1: No instruction sentence is printed for 問題1 in this recall retypeset — the page shows only the bare label 「問題1」, so the instruction field is left empty.
• 問題1: The reading-target words are enclosed in full-width parentheses (e.g. （錯覚）) instead of the official exam's underlining; transcribed as printed.
• 問題1: Question 5 stem reads 「記憶されている」, likely a recall/retype error for 「記録されている」; left as printed.
• 問題2: Q10 answer 1 is printed 堅苦しかった; given the context (何もできず、とても…) this may be a recall/retype error for 心苦しかった, but it is transcribed as printed.
• 問題2: Q13 stem ends （　　）しまった rather than （　　）てしまった; the て is supplied by the options (もたれて etc.) — transcribed as printed.
• 問題2: Q12 stem reads 習い始めると, which may be a recall-typo for 習い始めてから/習い始めると同時に; transcribed as printed.
• 問題3: The target words are printed enclosed in full-width parentheses, e.g. （リスク）, instead of being underlined as in official exams; transcribed exactly as printed.
• 問題3: In question 15 the stem uses アイディア (with small ィ) as printed.
• 問題3: In question 15 answer 4, the source prints 「問題だと厳しくしかった」 (likely intending 厳しくしかった = 叱った); transcribed as printed.
• 問題3: The printed 問題3 label has no instruction sentence after it (the retypeset omits the usual ＿＿＿の言葉に意味が最も近いものを… line), so the instruction field is empty.
• 問題3: The target word in each stem is printed inside full-width parentheses, e.g. （リスク）, instead of the official exam's underline; transcribed exactly as printed.
• 問題4: Headword styling is inconsistent across questions. In questions 22, 23, and 24 the headword occurrence appears in full-width parentheses and the WHOLE parenthesized group — both parentheses and the headword — is underlined (verified at high zoom: the underline starts under the opening （ and runs under the closing ）). In questions 20, 21, and 25 the parenthesized headword has no underline at all.
• 問題4: q21 option 3 prints 「昼食の時間になってので読書は…」, a suspected recall/retype typo for 「になったので」; transcribed as printed.
• 問題4: q22 option 2 begins 「個の書類のお名前の横に」, a suspected recall typo for 「この書類」; transcribed as printed.
• 問題4: q22 option 3 prints 「（押収）しまうので」, odd phrasing (possibly missing 「して」); transcribed as printed.
• 問題4: q25 headword is printed 「秘める」, but options 1 and 2 print the kanji as 「必めて」 (必 instead of 秘, verified at high zoom) while options 3 and 4 print 「秘める」 in unconjugated dictionary form; all transcribed as printed.
• 問題6: Question 37, option 2 is printed as 主人公 with no following particle (the official sentence would need 主人公を); transcribed as printed.
• 問題6: Question 40, option 3 is printed as 人手不足が深刻している, a likely recall typo for 深刻化している; transcribed as printed.
• 問題6: Question 40 option 3 is printed 人手不足が深刻している — likely a recall typo for 深刻化している; transcribed as printed.
• 問題6: Question 37 option 2 is printed 主人公 with no trailing particle — the genuine exam segment is believed to be 主人公を; transcribed as printed.
• 問題7: Paragraph 1 has unbalanced quotation brackets as printed: 「さっきから」ずっと…していない？」。 — a closing 」 right after さっきから plus a second closing 」 at the sentence end with no matching opener; likely a retype error (probably 「さっきから、ずっと…していない？」。). Transcribed as printed; the question mark prints half-width and is normalized to full-width ？ per convention.
• 問題7: 家じゆうに (paragraph 1) and 今日じゆうに (paragraph 5) are printed with full-size ゆ instead of small ゅ (じゅう); transcribed as printed.
• 問題7: Paragraph 4: half-width spaces are printed around 20 in 10分や 20 分; kept as printed (confirmed against the double-resolution renders).
• 問題7: Paragraph 5: 後 しして消耗を防ぎ is printed with a visible roughly half-width-space gap between 後 and しして and appears garbled (original possibly 後回しにして); transcribed with a single half-width space as printed.
• 問題7: Paragraph 6: そう考え、ある晩サボることにした出張から帰ったキャリーバッグを… has no punctuation between にした and 出張 (likely a dropped 。); transcribed as printed.
• 問題7: Paragraph 8 suspected retype errors, all transcribed as printed: 所定の地位置 (likely 定位置), 私にとつて with full-size つ (likely とって), and 付加より前にリセットだ (付加 looks like a recall/retype error but is clearly legible).
• 問題7: Question 43 option 2 is 片付ける限り, identical to question 42 option 2 and inconsistent with the three プロセス options — almost certainly a recall/copy error in the source; transcribed as printed.
• 問題8: None of the four passages has a printed source citation （…による）; the recall retype omits them.
• 問題8: Passage (1): 「相手方反論されることがあります」 as printed; likely a recall typo for 相手に/相手から反論される; transcribed as printed.
• 問題8: Passage (1): 「むしろブラスに解釈しましょう」 — ブ with dakuten clearly printed (verified at high zoom on p06-t); almost certainly a typo for プラス; transcribed as printed.
• 問題8: Passage (2): printed 「なぜ国によって全然違ってきます」 — likely missing か (なぜか国によって); transcribed as printed.
• 問題8: Passage (2) and question 46: 「美意義」 printed throughout (passage once; options 1, 2, 4); almost certainly a typo for 美意識; transcribed as printed.
• 問題8: Passage (2): printed 「工具も持つ美しさ」 — possibly a typo for 工具の持つ; transcribed as printed.
• 問題8: Question 46 options 1 and 3: 「合理的より優先」 and 「合理的の追求」 as printed; likely recall typos for 合理性; transcribed as printed.
• 問題8: Passage (3): 「起業社の才覚」 as printed; likely a typo for 起業者 or 起業家; transcribed as printed.
• 問題8: Question 47 option 2: 「人並びへた能力」 — the character between び and た is a full-size, baseline-aligned hiragana へ (verified at high zoom on p07-t), not a small raised caret; presumably a corrupted recall typo for 人並み外れた; transcribed as printed.
• 問題8: Passage (4): 「読者が存在しないビころか」 — katakana ビ clearly printed (verified at high zoom on p07-t); likely a typo for どころか; transcribed as printed.
• 問題8: Passage (4): 「これは限りないく私的な記録である筈の日記にとって」 as printed; likely a typo for 限りなく; transcribed as printed.
• 問題8: Passage (4) 注: 「他の人の日記や手紙など'をこっそり読む」 — a stray apostrophe-like tick is printed between など and を (verified at high zoom); transcribed as an ASCII apostrophe.
• 問題8: Question 48 references 矛盾である (「矛盾であるとあるが」), but no underline is printed in the passage or stem; the recall retype apparently lost the underline, so no <u> markup was added.
• 問題9: No 問題9 instruction sentence is printed in this recall reconstruction: the '問題9' label is immediately followed by passage（1）, so the 'instruction' field is left empty (consistent with how mondai8/mondai10 of this same source were handled).
• 問題9: Passage（1）contains apparent recall typos retained as printed: '妥協する余地を得るないが' (likely meant 妥協する余地はないが) and '自分の実力がまどうて"あれ' (garbled, likely 自分の実力がいまひとつであれ; a stray ASCII double-quote appears in the source).
• 問題9: Passage（2）recall typos retained as verified against the high-res images: 'ベテランのようなかをしている' (likely 顔をしている), 'こンは絶対に' (likely これは), '負荷や刺激を与える。' (sentence-ending 。 mid-sentence where 、 would be expected), 'たいだい30度' (likely だいたい), and '関節を柔軟いにし' (stray い; likely 柔軟にし). The page clearly prints these forms.
• 問題9: Q53 stem is a recall garble verified on the page: 'この文章のキーワードを「水温」の他に3つあげるとしたら、以下ビの3つるか。' — the character after 以下 is ビ (dakuten), not ピ; transcribed exactly as printed.
• 問題9: Q54 answer options are open-ended sentence fragments ending in '〜と' or '〜のが' (e.g. '...水中で行うと', '...合わせて行うのが'); transcribed exactly as printed.
• 問題9: Passage（3）: '略る' at end of first sentence is a garbled verb ending (likely おる/いる), transcribed as the closest glyph read. The word after 朗読CDは reads 趣味 ('朗読CDは趣味や実用に'), verified on the page (B's OCR base produced a spurious Cyrillic 'читал' here which the page does not support).
• 問題9: Passage（3）: 'それなりに、これが日本であまり利用されないのは', 'ほぼ同じ時間がかかる' and '理解できた内容もほぼ同じだというに違いない' all verified against p10; the leading 'ほぼ' and 'というに' are present on the page.
• 問題9: Passage（3）: 'ニュアンスの違いなど」が' (a stray closing 」 appears mid-sentence) retained as printed; likely a source artifact.
• 問題9: Q55 option 4 ends with a full-width 。 inside the option, unlike options 1-3; retained as printed.
• 問題9: Q56 option 1 is printed with NO leading number digit (it begins directly with '自身で黙読するのと音読するのとでは'); options 2-4 are numbered. Transcribed as printed.
• 問題10: Suspected retype glyph corruption (paragraph 2): 「根本的には①この傾向…」 is printed as ネ艮本的 — katakana ネ + 艮 as two separate, clearly spaced glyphs instead of 根. Confirmed against the double-res image. Transcribed as printed.
• 問題10: Suspected retype glyph corruption (paragraph 4): 「行為の根本を考究して」 is printed as 才艮本 — first glyph 才/オ (kanji 才 vs katakana オ are indistinguishable in this font; 才 chosen) followed by 艮, instead of 根. Elsewhere on the page 根 renders normally (成立根拠, 根本的な知), so these are literal characters in the retype, not a font artifact. Transcribed as 才艮本.
• 問題10: Foreign glyphs injected mid-word (paragraph 3): 「自分と無ĨIU関係な事柄」 — three solid Latin glyphs ĨIU (capital I-with-tilde U+0128, capital I, capital U) printed between 無 and 関係, baseline-aligned and same weight as the surrounding text, so they are part of the retype rather than the diagonal watermark (likely encoding corruption of 関 or branding injection). The first glyph is full cap height with a tilde, i.e. uppercase Ĩ, not lowercase ĩ. The visible gaps are automatic Latin/CJK spacing, not literal spaces, so transcribed without spaces.
• 問題10: Suspected recall/retype typo (paragraph 2 / opening): 「立場を我がのものとすることにある」 — 我がのもの instead of the expected 我がもの. Transcribed as printed.
• 問題10: Suspected recall/retype typo (paragraph 4): 「人間はいかに生きるであるかという」 — likely missing べき (いかに生きるべきであるか). Transcribed as printed.
• 問題10: Final paragraph prints 「こうした人生観: 世界観の根本的な知」 with an ASCII colon plus half-width space between 人生観 and 世界観 (original book likely had ＝ or ・). Same treatment applied to the colon in the 注 gloss, while the retype reading 「捨象（しゃしょう）」 is normalized to <ruby> markup in output.
• 問題10: Source citation is printed as 「（二 郎『現代の哲学』による）」 with a half-width space between 二 and 郎 — the author's family name appears mangled/lost in the recall version (the book 『現代の哲学』 exists). Transcribed exactly, gap as a half-width space.
• 問題10: Q60 stem ends 「…扱うことができないとあるがなぜか。」 with no comma before なぜか, as printed.
• 問題10: Q60 option 1: 「客観てきな事実確認」 — てき in hiragana, typo for 客観的な. Transcribed as printed.
• 問題10: Q61 option 2: 「主体的に考え方法を示せること。」 — reads oddly, likely missing る (考える方法). Transcribed as printed.
• 問題11: Half-width ASCII in the retype (digits, A&B, the standalone A/B letters) is printed with thin surrounding typesetting spaces (e.g. 「7 階建て」, 「8 月末」, 「A の手紙文」, 「B では」); these spaces are dropped per the digit/Latin normalization, while spaces between Japanese characters are kept as printed.
• 問題11: Passage A: printed 「貴店にもぜひご出店お願えればと存じます」 — 「お願えれば」 is likely a recall/retype error (for 願えれば / お願いできれば); transcribed as printed (verified at high zoom).
• 問題11: Passage B: printed 「現在の平屋になんじだ年配のお客様」 — なんじだ is presumably a garbled なじんだ; transcribed as printed (verified at high zoom).
• 問題11: Passage B: printed 「いずれ具体的なをことを直 9 相談申し上げたく」 — clearly garbled (presumably 具体的なことを直接ご相談); the stray を and the digit 9 are transcribed as printed, with the thin spaces around 9 dropped per the digit normalization.
• 問題11: Question 62 options 1 and 4: printed 「岸本,駅前」 with a half-width comma between 岸本 and 駅前 (stray retype artifact); option 4 also has a printed space before に (「駅前 に」); transcribed as printed.
• 問題11: Question 63 option 3: printed 「商店街にる現店舗よりも」 — likely missing あ (にある); transcribed as printed.
• 問題12: Suspected recall typo in the first sentence: 「周りの人のまなざいこ合わせるように」 (presumably まなざしに合わせる); transcribed as printed.
• 問題12: Suspected recall typo: 「多くの人々が生き交う街」 (presumably 行き交う); transcribed as printed.
• 問題12: Second paragraph has 「ひつようだからである」 in hiragana and the awkward double-が sentence 「それがはっきりと明文化されていったものが、法律や制度がたくさんあるように思える。」 — both transcribed exactly as printed. The 注 gloss 「（注）あげくの果てには：そして最後には」 confirms paragraph 2 reads あげくの果てには.
• 問題12: Fourth/fifth-paragraph print artifacts transcribed as-is: 「ルールが生まれるのてはなく」 (て without dakuten) and 「決められたルールにまさしを合わせる」 (reads まさし on the double-resolution image: missing な and no dakuten on さ; one transcriber read まざし, the other まなざし — the print is まさし).
• 問題12: Fifth paragraph: 「単なる方型にはまっていく」 — 方型 as printed, possibly a recall artifact for 鋳型/型.
• 問題12: Question 65 option 3 prints 「自然に身壬 についてくる」: a stray 壬 glyph after 身 followed by a visible space (presumably meant to be 身についてくる); transcribed with the space as printed.
• 問題12: Question 66 option 1 contains a stray interpunct: 「特に難しいこと・ではなくなってくる」; transcribed as printed.
• 問題12: Question 66 stem quotes 無意識の型となってくる with a printed underline (verified on the double-resolution image); kept as <u>…</u>. The corresponding phrase in the passage body is not underlined.
• 問題13: Suspected source/retype typo in the passage: the 〈正南ホールの使用〉 section begins 「艮請前に」 — almost certainly meant to be 「申請前に」 (the classroom section correctly prints 「申請可能な教室一覧」). The printed glyph is clearly 艮, not 申. Transcribed as printed.
• 問題13: Suspected source/retype typo in the second table, row A: 「参加者 5Q 名以上」 — the glyph is clearly a Latin Q (with a tail), almost certainly meant to be 「50」. Transcribed as printed (5Q).
• 問題13: Suspected source typo: 「まとめ提出してください」 — original was probably 「まとめて提出」 (missing て). Transcribed as printed.
• 問題13: Question 68 stem ends its first sentence with a Western-style dot 「思っている．」 instead of 「。」; transcribed as printed using full-width ．.
• 問題13: The retype's half-width spacing around ASCII tokens is inconsistent and was preserved exactly as printed: e.g. 「中教室102 - 106」 (no space after 室, row ①) vs 「中教室 202 - 206」 (space, rows ②–④), and 「3 か月前 / 2 か月前 / 7 日前」 (spaced) vs 「3日前」 (no space, row D).
• 問題13: First line of the document is a breadcrumb navigation line using half-width > separators (正南大学>在学生向け>…), transcribed as printed.
• 問題13: Availability column of the first table: the negative mark is a Latin-X-shaped glyph, transcribed as half-width X per the ASCII normalization rule; the positive mark is a circle transcribed as ○ (U+25CB) followed by full-width brackets ［各教室 1 台］ (printed with full-width 角 brackets and spaces around the 1), as printed.
• 問題13: In 「提出書類一覧（提出先 ： 学生課）」 the colon is printed as a full-width ： with a half-width space on each side; transcribed as printed.

Corrections applied (57)

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

• 問3 q18 question: 「<u>出馬</u>するらしい」 → 「<u>出馬する</u>らしい」
  • The choices are complete verb phrases, including the correct 選挙に出る. Leaving する outside the tested span produces 選挙に出るするらしい; the full target is 出馬する.
  • evidence: structural: 問題3 choices replace the underlined span.
• 問3 q19 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, as in the repeated N1 2014-07 q17 item.
  • evidence: structural: 問題3 choices replace the underlined span; repeated item in N1 2014-07 q17.
• 問4 q25 answers[0]: 「必めて」 → 「秘めて」
  • 必めて is a non-word; the 問題4 target word is 秘める (confirmed by the heading and answers[2]/[3], which read 秘める). The recall source mis-OCR'd 秘 as 必. One unambiguous intended word.
  • evidence: non-word + target-word constraint (秘める)
• 問4 q25 answers[1]: 「必めて」 → 「秘めて」
  • Same 必→秘 mis-OCR in the second distractor sentence of q25 (target word 秘める).
  • evidence: non-word + target-word constraint (秘める)
• 問1 q4 answers[3]: 「とど'めて」 → 「とどめて」
  • Stray ASCII apostrophe between ど and め; the word is とどめて.
  • evidence: freq+scribd witnesses both read とどめて (and 解析 glosses とどめて); yuuki-source-only apostrophe artifact. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問5 q30 question: 「言ったつけ」 → 「言ったっけ」
  • Full-size つ where sentence-final particle っけ requires small っ.
  • evidence: freq+scribd witnesses both read 言ったっけ. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問5 q30 question: 「兄「え？そんなこと言ったっけ？」 → 「兄「え？そんなこと言ったっけ？」」
  • The middle dialogue turn is missing its closing Japanese quote after the question mark; the surrounding 妹/兄/妹 turns are each quoted speaker turns.
  • evidence: freq witness lines 188-191 and the scribd retype both close the 兄 turn as 「言ったっけ？」. [paragraph normalization pass, 2026-07-02]
• 問5 q33 question: 「で'も」 → 「でも」
  • Stray apostrophe inside でも; the intended conjunction is でも.
  • evidence: freq+scribd witnesses read でも. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q41 passage: 「家じゆうに」 → 「家じゅうに」
  • Full-size ゆ where 家じゅう needs small ゅ.
  • evidence: freq+scribd witnesses both read 家じゅうに. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q42 passage: 「今日じゆうに」 → 「今日じゅうに」
  • Full-size ゆ where 今日じゅう needs small ゅ.
  • evidence: freq+scribd witnesses both read 今日じゅうに. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q44 passage: 「私にとつて」 → 「私にとって」
  • Full-size つ where とって needs small っ.
  • evidence: freq+scribd witnesses both read 私にとって. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問7 q41 answers[0]: 「驚くほど'だ」 → 「驚くほどだ」
  • Stray apostrophe inside 驚くほどだ.
  • evidence: freq+scribd witnesses (and 解析) read 驚くほどだ. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q47 answers[1]: 「人並びへた能力」 → 「人並び外れた能力」
  • へた is a garble for 外れた (人並み外れた = extraordinary); witnesses read 人並び外れた.
  • evidence: freq+scribd witnesses both read 人並び外れた能力 (only the yuuki source garbled 外れ→へ). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q48 passage: 「存在しないビころか」 → 「存在しないどころか」
  • カタカナ ビ is a garble for どころ.
  • evidence: freq+scribd witnesses both read 存在しないどころか (yuuki-source-only OCR error). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q48 passage: 「手紙など'を」 → 「手紙などを」
  • Stray apostrophe after など in the 注 gloss.
  • evidence: freq+scribd witnesses both read 手紙などをこっそり読む. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q49 passage: 「まどうて"あれ」 → 「いまどうであれ」
  • Garbled cluster with a stray ASCII double-quote for いまどうであれ.
  • evidence: freq+scribd witnesses both read 自分の実力がいまどうであれ. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q49 passage: 「妥協する余地を得るないが」 → 「妥協せざるを得ないが」
  • 余地を得るない is ungrammatical; context forces compromise on market cars.
  • evidence: freq+scribd witnesses both read 技術的に妥協せざるを得ないが (yuuki-source-only garble). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q49 passage: 「（注1）歯車：」 → 「（注1）歯ぎしり：」
  • The 注1 term must gloss the word used in the passage (歯ぎしりをする) and 「悔しく思う」 fits 歯ぎしり, not 歯車.
  • evidence: passage uses 歯ぎしり; freq+scribd witnesses gloss 歯軋り (= 歯ぎしり). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q52 passage: 「こンは絶対に」 → 「これは絶対に」
  • こン is a garble for これ.
  • evidence: freq+scribd witnesses both read これは絶対にまねてはいけない. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q53 question: 「以下ビの3つるか」 → 「以下のどの3つか」
  • Garbled recall question stem (以下ビの3つるか contains non-words ビ and つるか); best-judgment reconstruction of 「以下のどの3つか」 (which three of the following). Recall garble; the clean stem reads 以下どの3つを入れるか.
  • evidence: freq+scribd witnesses both read 以下どの3つを入れるか. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q55 passage: 「販売されて略る」 → 「販売されている」
  • 略る is a garbled verb-ending for いる.
  • evidence: freq+scribd witnesses both read ずらりと販売されている. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q55 passage: 「それなりに、これが日本」 → 「それなのに、これが日本」
  • それなりに does not fit the adversative contrast; それなのに is required.
  • evidence: freq+scribd witnesses both read それなのに、これが日本であまり利用されない. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q57 passage: 「違いなど」が」 → 「違いなどが」
  • Stray closing 」 mid-sentence.
  • evidence: freq+scribd witnesses both read ニュアンスの違いなどが. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q57 passage: 「「間」など、を含めた」 → 「「間」などを含めた」
  • Stray comma between など and を.
  • evidence: freq+scribd witnesses both read 「間」などを含めた音声表現. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q58 passage: 「我がのもの」 → 「我がもの」
  • Extra の; the idiom is 我がものとする.
  • evidence: freq+scribd witnesses both read 立場を我がものとすることにある. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q59 passage: 「ネ艮本的には①」 → 「根本的には①」
  • Glyph corruption: katakana ネ + 艮 for 根.
  • evidence: freq+scribd witnesses both read 根本的には①この傾向. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q60 passage: 「無ĨIU関係な」 → 「無関係な」
  • Injected Latin glyph cluster ĨIU inside 無関係.
  • evidence: freq+scribd witnesses both read 自分と無関係な事柄. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q61 passage: 「行為の才艮本を考究」 → 「行為の根本を考究」
  • Glyph corruption: 才(オ) + 艮 for 根.
  • evidence: freq+scribd witnesses both read 人間の主体的な行為の根本を考究して. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問11 q62 answers[0]: 「岸本,駅前の」 → 「岸本駅前の」
  • Stray comma inside the place name 岸本駅前 (used cleanly in passageA).
  • evidence: passageA reads 岸本駅前 throughout with no comma. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問11 q62 answers[3]: 「岸本,駅前 に新しくできるビルに1若者層」 → 「岸本駅前に新しくできるビルに、若者層」
  • Stray comma and space in 岸本駅前, plus a stray digit 1 where a comma belongs before 若者層.
  • evidence: passageA reads 岸本駅前; OCR witnesses both show に新しくできるビルに1若者層を呼び込む (the 1 is a stray glyph for the comma). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q65 passage: 「まなざいこ合わせる」 → 「まなざしに合わせる」
  • Garble for まなざしに合わせる.
  • evidence: OCR pass B reads 周りの人のまなざしに合わせるように (OCR pass A garbled it); matches Google Books まなざしのデザイン. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q66 passage: 「まさしを合わせる」 → 「まなざしを合わせる」
  • まさし is a garble for まなざし (used throughout the passage).
  • evidence: OCR pass B reads 決められたルールにまなざしを合わせる (OCR pass A garbled it). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q66 passage: 「生まれるのてはなく」 → 「生まれるのではなく」
  • て without dakuten in のではなく.
  • evidence: OCR pass B reads 生まれるのではなく (OCR pass A garbled it). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q65 answers[2]: 「自然に身壬 についてくる」 → 「自然に身についてくる」
  • Stray glyph 壬 and space inside 身につく.
  • evidence: OCR witness shows 自然に身壬 についてくる; idiom is 身につく. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q68 passage: 「艮請前に」 → 「申請前に」
  • 艮請 is a glyph corruption of 申請 (the classroom section prints 申請可能な教室一覧).
  • evidence: freq+scribd witnesses both read 申請前に. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問13 q68 passage: 「参加者 5Q 名」 → 「参加者 50 名」
  • 5Q is a Latin-Q corruption of 50.
  • evidence: freq+scribd witnesses both read 参加者50名以上. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問1 q5 question: 「克明</u>に記憶されている」 → 「克明</u>に記録されている」
  • 克明に記録する is the natural collocation for documenting detail in a 文書; 記憶される (be remembered) does not fit 「文書には…されている」. Likely a recall/retype slip 記録→記憶.
  • evidence: Category homophone, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問4 q21 answers[2]: 「昼食の時間になってので」 → 「昼食の時間になったので」
  • なってので is an impossible form; the past tense なったので is required. This is an unrelated typo, not the deliberate リタイア misuse that makes the distractor wrong.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問4 q22 answers[1]: 「個の書類のお名前の横に」 → 「この書類のお名前の横に」
  • 個の書類 is ungrammatical; 「この書類」is intended. A typo independent of the 押収 misuse that makes the distractor wrong.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問6 q40 answers[2]: 「人手不足が深刻している」 → 「人手不足が深刻化している」
  • 深刻 is a na-adjective and cannot take する directly; 深刻化する (become serious) is the intended verb. Clear dropped 化.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問7 q41 passage: 「モノを所定の地位置に戻すのは」 → 「モノを所定の定位置に戻すのは」
  • 地位置 is a non-word; 所定の定位置 (designated fixed position) is the intended phrase. Retype duplicated/substituted 地 for 定.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q45 passage: 「むしろブラスに解釈しましょう」 → 「むしろプラスに解釈しましょう」
  • ブラス is a non-word here; プラスに解釈する (interpret positively) is intended. カタカナ濁点 typo ブ for プ.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q46 passage: 「民族独特の美意義の違いがくっきり」 → 「民族独特の美意識の違いがくっきり」
  • 美意義 is a non-word; the intended term is 美意識 (aesthetic sense). Retype 識→義 (visually/phonetically close).
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q46 answers[0]: 「民族独特の美意義が、合理的より優先」 → 「民族独特の美意識が、合理性より優先」
  • 美意義→美意識 (non-word); 合理的より→合理性より, since the noun 合理性 (rationality) is what is being contrasted, not the adjective 合理的.
  • evidence: Category typo, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q46 answers[1]: 「民族によって異なる美意義が」 → 「民族によって異なる美意識が」
  • 美意義 is a non-word for 美意識 (aesthetic sense).
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q46 answers[2]: 「国による違いは、合理的の追求から」 → 「国による違いは、合理性の追求から」
  • 合理的の is ungrammatical (な-adjective + の); the intended noun is 合理性 (合理性の追求 = pursuit of rationality).
  • evidence: Category typo, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q46 answers[3]: 「民族独特の美意義が反映されている」 → 「民族独特の美意識が反映されている」
  • 美意義 is a non-word for 美意識 (aesthetic sense).
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q47 answers[1]: 「人並び外れた能力を」 → 「人並み外れた能力を」
  • 人並び外れた is a non-word; the idiom is 人並み外れた (extraordinary). Retype び for み.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問8 q48 passage: 「これは限りないく私的な記録である筈」 → 「これは限りなく私的な記録である筈」
  • 限りないく is impossible; the adverb is 限りなく. The same passage opens with 「限りなく私的な記録」, confirming the intended form.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q52 passage: 「プールの水温はたいだい30度前後です」 → 「プールの水温はだいたい30度前後です」
  • たいだい is a non-word; だいたい (approximately) is intended. Mora transposition typo.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問9 q54 passage: 「あわせて関節を柔軟いにし」 → 「あわせて関節を柔軟にし」
  • 柔軟いに is ungrammatical (柔軟 is a na-adjective); the stray い should be dropped to give 柔軟にし.
  • evidence: Category typo, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問10 q60 answers[0]: 「客観てきな事実確認」 → 「客観的な事実確認」
  • 客観てき should be 客観的 (the passage elsewhere uses 客観的); kana てき is a retype slip for the kanji 的.
  • evidence: Category orthography, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問10 q61 answers[1]: 「主体的に考え方法を示せること。」 → 「主体的に考える方法を示せること。」
  • 考え方法 needs the verb 考える to modify 方法 (考える方法 = a way to think); dropped る.
  • evidence: Category dropped-char, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問10 q58 passage: 「人間はいかに生きるであるかという」 → 「人間はいかに生きるべきであるかという」
  • いかに生きるであるか is ungrammatical; parallels the earlier 「いかに生きるべきかをよく考えて」, so べき was dropped (いかに生きるべきであるか).
  • evidence: Category dropped-char, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問11 q62 passageB: 「商店街にる現店舗よりも」 → 「商店街にある現店舗よりも」
  • にる is impossible; the relative clause needs ある (商店街にある現店舗 = the current shop located in the shopping street). Dropped あ.
  • evidence: Category dropped-char, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問12 q65 passage: 「多くの人々が生き交う街」 → 「多くの人々が行き交う街」
  • 行き交う (いきかう, to come and go / cross paths) is the established word for streets full of people; 生き交う is a non-word homophone slip.
  • evidence: Category homophone, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]
• 問13 q68 passage: 「提出期限の最も早い書類に合わせ、まとめ提出してください」 → 「提出期限の最も早い書類に合わせ、まとめて提出してください」
  • まとめ提出 needs the て-form 連用 to connect: まとめて提出する (submit together). Dropped て.
  • evidence: Category dropped-char, confidence high. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]