N1 2023-07 — audit report

66 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 re-typeset digital text throughout (5-level legibility), but every page bears a large diagonal 'Tôi Yêu Ngoại Ngữ Group / Yuuki Bùi' watermark that overlaps text, and re-typing introduces occasional typos (同のように for どのように in Q60), mixed full-/half-width digits, kana substitutions in listening options, and a watermark-corrupted footer line on the page-30 flyer. Three pages are unreadable gray dither filler (6, 13, 31) but carry no content.

Every section was transcribed twice independently (a full-page pass and a double-resolution half-page pass) and reconciled. Of 13 sections, 7 agreed on all content between the two passes (reconciled automatically, differing only in transcriber notes) and 6 had at least one content difference resolved by a third adjudication pass that re-read the page images. 87 leaf-level differences were examined in total.

Residual OCR risks (25):

• 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: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題3: The Q16/Q19 corrections override values both transcribers agreed on; they rest on pixel measurement of underline extent, calibrated against the four other questions on the same page where the underline consistently overshoots glyph edges by only 2-4px, so confidence is high but they were not independently transcribed that way by either pass.
• 問題3: Q14 「この条約は締結されると」 retained exactly as printed even though が would be the expected particle; if the upstream retypeset introduced this, the official exam text may differ.
• 問題4: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題5: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 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: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題8: The intended correction for the 「後き」 typo in passage (2) is uncertain (likely 「浅き」/「浅さ」); transcribed as printed and flagged only.
• 問題8: Spacing around the long dash 「走ってしまう ―― 要は」 is printed with typographic padding; both transcribers (and the final) render it as 走ってしまう――要は with no spaces around the ――, treating the padding as glyph spacing rather than literal spaces. The inline mid-word space before 走って is retained as a genuine stray space.
• 問題8: Absence of any source citation （…による） across all four passages is atypical for genuine JLPT 問題8 and likely an artifact of the Vietnamese retypeset; nothing to transcribe but worth noting for downstream validation.
• 問題9: Q52 option 3 stray space rendered as a half-width space; the printed gap is close to a full-width cell, so it could arguably be a full-width 　 instead. Both transcribers read it as a space; width is the only uncertainty.
• 問題9: Passage (2)/(4): I treat side-bearing whitespace around half-width digits as non-content (no spaces), consistent with N1_2023-12; if the project later decides to preserve them, passage (2) digit runs and the instruction would need spaces.
• 問題9: Several printed retyping typos (問うで, 種は熟すのは, 同じじょう, 問題性 vs 問題点, missing 。 after 改変する) are transcribed as printed per convention, not corrected.
• 問題10: The half-width spaces kept in 「300 以上」, 「700 以上」, and 「きっ かけ」 are a judgment call: the gaps are clearly wider than normal kerning so they were kept as typed spaces per transcribe-as-printed, but a retyped Word document's automatic Latin/CJK spacing could in principle produce similar gaps for the digit cases.
• 問題10: The diagonal watermark crosses paragraphs 2-4 on p23-t; プログミング sits near the watermark but was confirmed at high zoom, leaving little doubt.
• 問題10: Several odd phrasings (動作プログラムをし、その結果複雑に、一から、すべては) are retained as printed; they look like retyping omissions from the original exam text and may differ from the genuine exam wording.
• 問題11: The gap after the full-width colon in the 注 gloss could conceivably be a typed half-width space in the retypeset; both transcribers and the adjudicator read it as glyph spacing of ：, so no space was transcribed.
• 問題11: The gaps around the letters A/B throughout (e.g. 「次の A と B の」) were treated as full-width-glyph spacing and dropped during half-width normalization; if the source actually typed half-width letters with literal spaces, an exact-as-printed transcription would include those spaces. The chosen treatment matches both transcribers and the N1_2023-12 pilot.
• 問題11: The space between the digit 4 and から in the instruction was kept as a typed space (alongside the unambiguous kana-kana spaces); if the 4 is a full-width glyph, part of that gap could be glyph spacing rather than a space.
• 問題12: no human/agent spot-check was performed for this section (A/B byte-agreement on content was treated as sufficient)
• 問題13: The CID→Latin junction gaps (ダンス B, 定員 20 名, づくり YuukBui) are 3.0pt glyph offsets, not space characters, in the source PDF (3.6pt is a true space); they were transcribed as spaces because they print as such, but a text-layer-faithful transcription would omit them.
• 問題13: The footer’s wide alignment gap between センター and 電話： has no convention; it was collapsed to a single full-width space, which is a judgment call.
• 問題13: The watermark-corrupted footer (YuukBui spliced into the center name, dubious phone number 02-1124-45) means the genuine printed footer of the real exam is unknown; transcription reflects this retypeset source only.
• 問題13: This retypeset source omits the standard 問題13 instruction sentence; the real exam almost certainly had one, so the empty instruction field reflects the source, not the original 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) — 71 noted at OCR time:

• 問題2: Question 7 stem reads 「暮らしているが卒業して」 with no comma between the clauses, as printed.
• 問題2: Question 13 answer 1 is printed as 行き詰って (no っ after 詰, i.e. not 行き詰まって), transcribed as printed.
• 問題2: Question 7: printed with no comma between 暮らしているが and 卒業して (「暮らしているが卒業して就職したら、」); transcribed as printed.
• 問題2: Question 13 option 1: printed as 行き詰って (without ま, i.e. not 行き詰まって); transcribed as printed.
• 問題3: Question 14 reads 「この条約は締結されると」 with は where が would be more natural before the passive 締結される; transcribed as printed (possible retyping artifact, but also plausible as a topicalized construction).
• 問題3: Underline extents as printed, confirmed by pixel measurement against the double-resolution renders: 14 懸念, 15 やつれて, 16 奮闘して, 17 不慮の, 18 根こそぎ, 19 没頭して. Both transcribers had stopped 16 and 19 one character short (奮闘し／没頭し); the printed underline clearly includes て in both, consistent with the て-form answer options.
• 問題4: Question 20 option 1 reads 「来月は100万台バイクを生産する兆しです。」 as printed; 「100万台バイク」 (rather than 「100万台のバイク」) looks like a retyping artifact but is transcribed as printed.
• 問題4: In question 22 the underline covers exactly さえて in all four options (verified at high zoom; in options 1 and 4 the following い is not underlined).
• 問題4: The retypeset source renders half-width digits and Latin letters with visible gaps from adjacent kana/kanji (e.g. 「100 万台」「A チーム」「8 万語」「3 年間」「2 時間」); these look like word-processor auto-spacing rather than typed spaces, so no spaces were transcribed around digits or Latin letters.
• 問題5: Question 33: the stem line ends with a closing 」 but has no matching opening 「 anywhere in the question (the line after （ホームページで） starts directly with 先日の…); likely a retypesetting artifact of an original quoted notice, transcribed as printed.
• 問題5: Question 33: the stem ends with a closing quote 」 that has no matching opening quote in the printed source — likely a retyping artifact; transcribed as printed.
• 問題6: Worked example on p09 has retyping typos: 「正しい分はこうです」 (分 for 文) and 「（解答の仕方）」 (仕方 for しかた); the example block is skipped per conventions so these do not appear in the data.
• 問題6: Question 40: the print shows visible gaps around the half-width digits（「人口 10 万人」「30 年前」style spacing）; transcribed without the spaces, treating them as digit-kerning of the retypeset rather than deliberate spacing.
• 問題6: The skipped worked-example block (解答の仕方) contains the source typo 「正しい分はこうです」 (分 for 文); not part of the transcribed questions.
• 問題7: Suspected retyping typo: 「書き直せば良さそうなものだが。どうしてもできない。」 has a full stop after 「ものだが」 where a comma would be expected; transcribed as printed.
• 問題7: Suspected retyping typo: 「その間いに得たものも多いが」 — 「間いに」 is likely an error for 「間に」; transcribed as printed.
• 問題7: Question 42's options are printed with visible half-width spaces inside the phrases (「褒めたり 叱ったり…」 and, in option 4, 「…しても 仕方ない」); transcribed as printed.
• 問題7: The retype's 注 glosses give readings in narrow half-width parentheses, e.g. 倦(あぐ)んだ, in contrast to the full-width （注1） labels; normalized to <ruby> markup in output.
• 問題7: Suspected retyping typo in the paragraph containing 【43】: 「その間いに得たものも多いが」 — 間い verified at high zoom (門 with 日 inside, followed by い); almost certainly meant 「その間に」. Transcribed as printed.
• 問題7: The 注 gloss lines print the readings in half-width ASCII parentheses, e.g. 倦(あぐ)んだ; normalized to <ruby> markup in output, with the full-width ： as printed. The layout tab gap after each （注N） label is dropped, matching the convention example and prior finals.
• 問題7: Question 42's options contain typed half-width spaces in the retypeset: after 褒めたり in all four options, and also before 仕方ない in option 4; verified at high zoom and transcribed as printed.
• 問題8: None of the four passages has a printed source citation （…による） or 注 glosses, unusual for JLPT 問題8; presumably dropped in the Vietnamese retypeset.
• 問題8: Q45 option 2: stray half-width space printed inside the word, 「情報を得るこ とに熱心な人は…」 — transcribed as printed.
• 問題8: Passage (2): stray half-width space printed in 「自分の文章 を読みながら」 — transcribed as printed.
• 問題8: Passage (2): 「『思ったこと』や『考えたこと』の後きや甘さを」 — 「後き」 is almost certainly a retyping typo (likely intended 「浅き」/「浅さ」); transcribed as printed.
• 問題8: Passage (3): stray half-width space printed in 「新奇性ばっかりに 走ってしまう」 — transcribed as printed; the following long dash normalized to ―― (two U+2015) per conventions (printed with surrounding typographic spaces, not transcribed).
• 問題8: Passage (3): 「いかがなものでしょうか？本質」 uses a full-width ？ (verified at high resolution), kept full-width per conventions.
• 問題8: Q48 option 1: stray half-width space printed before the comma, 「違いが大きいほど 、熱い対話ができる」 — transcribed as printed.
• 問題8: Instruction line prints half-width-looking spaces around 1・2・3・4 (「ものを 1・2・3・4 から」); treated as typography of half-width digits and not transcribed as spaces.
• 問題9: Passage (1): printed 「問うで学ぶ」 (verified at high zoom) — almost certainly a retyping typo for 「問うて学ぶ」; transcribed as printed.
• 問題9: Passage (1): printed 「常識を疑ってみる」ということ、実は… (no で after ということ) and 「常識とされてきたものの見方」, 「学ぶこととは」, 「ましてや」, 「自分の物差しが変わり、自分自身が変わっていく」 — all verified against the double-resolution p18-t.
• 問題9: Passage (2): printed 「カタクリの種は熟すのは花が咲いてから」 (verified) — suspected retyping typo for 「種が熟すのは」; transcribed as printed.
• 問題9: Passage (2) Q52 option 3: a genuine extra space is printed between 「アリの」 and 「死んだ」 (clearly wider than normal kerning, on a non-justified line); kept as a half-width space as printed.
• 問題9: Passage (3): printed 「みんなと同じじょうにできる」 (verified) — suspected retyping typo for 「同じように」; transcribed as printed.
• 問題9: Passage (3): body reads 「…生活全体のなかで、今全く新しく意味付けられた…」 — the 今 is present (verified).
• 問題9: Passage (3): note marker 3 is printed 「（注 3）」 with a genuine space (注1/注2 are tight), both inline and in the note line; kept as printed. Note-gloss readings are in half-width parens in the retype, e.g. 如何(いか)にも, 揶揄(やゆ)される, 見境(みさかい)なく, and normalized to <ruby> markup in output.
• 問題9: Passage (4): printed 「…という問題性を孕んでいる」 in the body (verified), while the Q55 stem reads 「問題点」 — a genuine source inconsistency; each transcribed as printed.
• 問題10: Instruction line prints the choice list as 「1．2・3・4」 — a full-width period after the 1 instead of the usual ・ separator; transcribed as printed (digits normalized to half-width).
• 問題10: Confirmed retyping typo in paragraph 2: 「ロボットに対するプログミングは」 — almost certainly meant プログラミング (paragraph 4 spells it correctly); verified at high zoom in the double-resolution render and transcribed as printed.
• 問題10: Paragraph 3 has 「300 以上」 and 「700 以上」 with a visible half-width-space-sized gap between the digits and 以上, clearly wider than Latin/CJK kerning elsewhere; adjudicated as an actual typed space in the retyped source and kept as printed.
• 問題10: Paragraph 3 phrase 「300 以上の動作プログラムをし、」 reads oddly (a word like 用意 may have been dropped in retyping); transcribed as printed.
• 問題10: Paragraph 3 sentence 「その結果複雑に、多様に動くロボットが実現された。」 has odd comma placement (likely meant その結果、複雑に); transcribed as printed.
• 問題10: Paragraph 5 contains a stray half-width space inside a word: 「何かをきっ かけに作動した」; the gap is clearly visible at high zoom and kept as printed.
• 問題10: Paragraph 7 phrase 「相手のことを一から、すべては理解しきれない」 reads oddly (possible retyping omission); transcribed as printed.
• 問題11: Question 60 stem is printed 「ＡとＢは同のように述べているか」 — 「同のように」 is almost certainly a retyping typo for 「どのように」; transcribed as printed.
• 問題11: The instruction line contains stray typed spaces in the retypeset between kana words (「最もよい ものを」, 「4 から 一つ選びなさい」), confirmed at high zoom; transcribed as printed (with digits/letters normalized to half-width ASCII).
• 問題11: Passage A, second paragraph: 「ヒントを得たいといつも問題意識を持っていれば」 reads awkwardly (a comma after 得たいと may have been dropped in retyping); transcribed as printed.
• 問題11: The letters A and B and adjacent gaps: one transcriber read full-width Ａ/Ｂ (gap = glyph spacing), the other half-width letters with surrounding retypeset spaces; either way the adjudicated output is half-width ASCII with no surrounding spaces, consistent with the pilot's treatment.
• 問題12: Instruction line: the first separator in the choice list is printed as a baseline period, 「1．2・3・4」 (verified at 6x zoom), not ・; transcribed as printed with a full-width ．(digits normalized to half-width ASCII).
• 問題12: Paragraph 1: 「不安な時代である、そしてそんな…」 — a comma is printed after である where a 。 would be expected (verified at zoom); suspected retyping error; transcribed as printed.
• 問題12: Paragraph 2: stray half-width space printed inside the word in 「教育で解決し ようというのは」; transcribed as printed.
• 問題12: The note-3 reference is printed with a stray space, 「（注 3）」, both in the body (躍起になっても（注 3）、) and in the gloss line, while 注1 and 注2 have no space; transcribed as printed with a half-width space.
• 問題12: Final paragraph: 「この社会と社会と教育を持続させる」 — 「社会と」 appears duplicated; suspected retyping error; transcribed as printed.
• 問題12: No source citation （…による） is printed at the end of the passage, unusual for JLPT 問題12; presumably dropped in the Vietnamese retypeset.
• 問題12: Q62 option 3: 「子ども一人一人に併せて」 — 併せて is likely a retyping variant of 合わせて; transcribed as printed.
• 問題12: Q63 option 3 lacks the sentence-final 。 unlike the other three options; transcribed as printed.
• 問題12: The passage mixes 子供 (paragraph 1, and 子供の方 in paragraph 3) and 子ども (elsewhere); transcribed as printed.
• 問題12: Instruction line prints the choice list as 「1．2・3・4」: the separator after 1 is a baseline full-width period, not a middle dot — suspected retyping typo, transcribed as printed.
• 問題12: First paragraph: 「…不安な時代である、そしてそんな…」 uses 、 after である where 。 would be expected — suspected retyping typo, transcribed as printed.
• 問題12: Third paragraph: 「きっちりきっかり」 as printed (standard phrase would be きっちりかっきり) — suspected retyping typo.
• 問題12: 「（注 3）」 is printed with a space between 注 and 3, both at the inline reference and in the gloss line, unlike （注1）／（注2） — preserved as printed.
• 問題12: Final paragraph: 「この社会と社会と教育を持続させる」 — the doubled 社会と looks like a retyping duplication; transcribed as printed.
• 問題12: No source citation line (（…による）) is printed after the passage in this retypeset.
• 問題12: Question 62 option 3 prints 併せて where 合わせて is expected — suspected retyping typo, transcribed as printed.
• 問題12: Question 63 option 3 has no final 。 as printed (options 1, 2, 4 end with 。).
• 問題13: The markers before 基本予約 and 追加予約 are the glyph ✙ (U+2719 outlined Greek cross, Segoe UI Symbol in the source); kept as printed since it is not covered by the normalization rules.
• 問題13: The 〈時間〉 content line prints 10:40-1２:00 with a mixed half-width 1 and full-width ２; normalized to 12:00 per the digit conventions. The range separator and the phone-number dashes are ASCII hyphens as printed. The line has printed spaces before 〈会場〉 and 〈料金〉 but none after the closing brackets.
• 問題13: (通年開講) is printed with half-width ASCII parentheses, unlike the full-width parentheses used elsewhere; transcribed as printed.
• 問題13: In the class table, the ダンスA row’s 内容 cell ends without 。 while the other two rows end with 。; transcribed as printed (suspected retyping omission).
• 問題13: Retypeset spacing artifacts kept as printed: 「次の 4 人」 (spaces around 4) in Q65; 「54 歳」 in the Q65 table while 34歳/16歳/27歳 have no space; 「ダンス B」 in the class table vs 「ダンスA」 in the row above, plus 「ダンス A」 in Q66; （各クラス定員 20 名）; and a space after 電話：. At the ス→B, 員→20, and り→YuukBui junctions the source has a space-width glyph offset rather than a space character, but it prints as a space, so a space is transcribed.
• 問題13: The document’s final line embeds the retypesetter’s watermark name in the body text: 木原市健康づくり YuukBui センター (the genuine text is presumably 木原市健康づくりセンター), and the phone number 02-1124-45 is likewise suspect; transcribed as printed. The left-aligned center name and the right-aligned 電話： 02-1124-45 sit on one printed line; the wide alignment gap was collapsed to a single full-width space.

Corrections applied (10)

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

• 問7 q43 passage: 「その間いに得たものも多いが」 → 「その間に得たものも多いが」
  • 「間い」 is a non-word retype artifact; intended word is 「間（あいだ）」. Published source confirms 「その間に得たものも多いが喪った」.
  • evidence: Google Books: 浅田次郎 — snippet 「その間に得たものも多いが喪ったものもまた多かろう」; retype PDF line 459 shows the artifact. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問8 q46 passage: 「の後きや甘さを」 → 「の浅さや甘さを」
  • 「後き」 is a non-word. Parallel with 「甘さ」 (さ-nominalized adjective) and context (思い知らされる — made aware of shallowness/naivety) require 「浅さ」.
  • evidence: Non-word in retype PDF line 579; grammatical parallel with 「甘さ」 in same phrase. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q49 passage: 「学問は「問うで学ぶ」ことであり」 → 「学問は「問うて学ぶ」ことであり」
  • 「問うで」 is a non-word; te-form of 問う is 問うて. Parallels 「強いて勉める」 in the prior clause.
  • evidence: te-form of 問う is 「問うて」, never 「問うで」; retype PDF line 700 shows the で/て artifact. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q51 passage: 「カタクリの種は熟すのは花が咲いてから」 → 「カタクリの種が熟すのは花が咲いてから」
  • Double 「は」 (種は…のは) is ungrammatical; subject of subordinate clause 熟す should take が.
  • evidence: Grammar: double topic は is ungrammatical; single は→が character substitution; retype PDF line 739. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問9 q53 passage: 「みんなと同じじょうにできること」 → 「みんなと同じようにできること」
  • 「じょうに」 is a kana retype error for 「ように」 (じょ/よ confusion); 「同じように」 is standard.
  • evidence: Kana typo in retype PDF line 785; 「同じように」 is the standard expression. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問10 q57 passage: 「ロボットに対するプログミングは」 → 「ロボットに対するプログラミングは」
  • 「プログミング」 is a retype omission of ラ; the word is 「プログラミング」.
  • evidence: Same passage spells プログラミング correctly (retype PDF line 915) and プログラム throughout (lines 911-958). [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問11 q60 question: 「AとBは同のように述べているか」 → 「AとBはどのように述べているか」
  • 「同のように」 is a non-word retype error for 「どのように」, the standard stem wording.
  • evidence: Same exam Q48 uses 「どのように述べているか」 (retype PDF line 662); Q60 artifact at line 1046. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q62 passage: 「この社会と社会と教育を持続させる」 → 「この社会と教育を持続させる」
  • 「社会と」 is duplicated; intended phrase is 「この社会と教育を持続させる」.
  • evidence: Clear duplication in retype PDF line 1118. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q62 answers[2]: 「子ども一人一人に併せて」 → 「子ども一人一人に合わせて」
  • 「〜に合わせて」 (adapt/match to) uses 合わせる; 併せる (combine) does not take に in this sense. Wrong-kanji retype error.
  • evidence: Internal consistency: the same exam uses 合わせて for に-marked adapt-to in three other places — 「能力に合わせて」(line 865), 「目的に合わせて」(line 1145, adjacent answer 63-1), 「リズムに合わせて」(line 1295); only this answer uses 併せて. [correct-retypes workflow (propose + adversarial verify), 2026-06-17]
• 問12 q62 passage: 「「必要な人材をきっちりきっかり、耳をそろえて」 → 「「必要な人材をきっちりかっきり、耳をそろえて」
  • The conventional emphatic reduplication is 「きっちりかっきり」 (both meaning exactly/precisely, paired). 「きっちりきっかり」 repeats きっ at the head of the second element, which is not the idiomatic form; the second element should be かっきり. Likely a retype transposition.
  • evidence: Category typo, confidence medium. [retype-corrections-pass2 (propose + adversarial verify), 2026-06-20]