chaffjs 0.30.0 → 0.32.0

package/rules/name-variant.yaml

@@ -21,15 +21,24 @@ messages:
   spelling:
     ja: 「{name}」は、ほかの所では「{usual}」と書いています（字の大小・幅・記号の違い）
     en: '"{name}" is written "{usual}" elsewhere in the document (case, width or punctuation)'
+  kana:
+    ja: 「{name}」は、ほかの所では「{usual}」と書いています（名前の中の、同じ音を書く字の違い。ヶ と ケ など）
+    en: '"{name}" is written "{usual}" elsewhere in the document (the same sound written with another character inside the name)'
   reading:
     ja: 「{name}」は、ほかの所では同じ読みの「{usual}」と書いています
     en: '"{name}" is written "{usual}", read the same, elsewhere in the document'
   character:
     ja: 「{name}」は、ほかの所では字体の違う同じ字で「{usual}」と書いています
     en: '"{name}" is written "{usual}", with another form of the same character, elsewhere in the document'
+  lookalike:
+    ja: 「{name}」は、ほかの所で何度も書いた「{usual}」と、形の似た字一字だけが違います
+    en: '"{name}" differs from "{usual}", which the document uses more than once, by one similar-looking character'
   near:
     ja: 「{name}」は、ほかの所で何度も書いた「{usual}」と一字違いです
     en: '"{name}" is one letter away from "{usual}", which the document uses more than once'
+  field:
+    ja: 「{name}」は、人の名前を書く欄（氏名：など）では「{usual}」と書いています
+    en: '"{name}" is written "{usual}" in the field that names the person (Name:, Employee: and the like)'
   surname:
     ja: 「{name}」は、ほかの所で何度も書いた「{usual}」と名が同じで、姓が数字だけ違います
     en: '"{name}" has the given name of "{usual}", which the document uses more than once, and a surname a few letters away'
@@ -48,6 +57,12 @@ messages:
   company-near:
     ja: 「{name}」は、ほかの所で書いた会社名「{usual}」と一字違いです
     en: '"{name}" is one letter away from the company name "{usual}" written elsewhere in the document'
+  company-kana:
+    ja: 「{name}」は、ほかの所では「{usual}」と書いています（名前の漢字一字を、その読みのかなで書いた違い）
+    en: '"{name}" is written "{usual}" elsewhere in the document (one kanji of the name written in kana)'
+  model-code:
+    ja: 「{name}」は、ほかの所ではハイフン・空白・字の大小を変えて「{usual}」と書いています
+    en: '"{name}" is written "{usual}", with other hyphens, spaces or case, elsewhere in the document'
 
 how_to_fix:
   ja: 正しい形を確かめ、一つに揃えてください。本当に別の人や会社なら、そのままで構いません。
@@ -68,7 +83,7 @@ use_for: [business, technical, blog, legal, docs, academic, literature, speech]
 group: consistency
 summary:
   ja: 同じ人・会社・製品の名前を、少しだけ違う形で書いている所（山田太郎 と 山田太朗、GitHub と Github）
-  en: The same person, company or product written in slightly different forms (GitHub and Github, Acme, Inc. and Acme Inc.)
+  en: The same person, company or product name written slightly differently
 example:
   ja:
     before: |
@@ -82,8 +97,8 @@ example:
       The code lives on GitHub. Reviews happen on GitHub, and releases are tagged on GitHub.
 
 not_flagged:
-  ja: 品詞解析が固有名詞と読んだ語だけを見ます。大文字だけの名前（ACME INC、NASA）は見ません。一字違いの英字の名前は、6 字以上で、多いほうが二度以上、少ないほうが一度だけのときに限ります（Iran と Iraq は別の名前）。人の名前どうしは、字体の違う同じ字（斎藤 と 斉藤、渡辺 と 渡邊）か、読みが同じで一字だけ違い、多いほうが二度以上、少ないほうが一度だけのとき（松本 と 松元）に限ります（伊藤 と 伊東 は別の人のことがある）。人の名前とは、品詞解析が人名と読んだ語か、「様」「さん」「氏」の付いた名前です（解析器が名前と読めない字体の字を含む名前は、敬称の前の 2〜4 字の漢字を名前と読みます）。敬称の無い名前も、前後に漢字が続かない 2〜4 字の漢字で、片方の書き方に「担当の」「私、」などの前置きか敬称が付き、もう片方が「です」「が」「までご連絡」などの前にあれば比べます（担当の斎藤です と 斉藤までご連絡ください）。地名にも同じ組がある（鹿島 と 鹿嶋）ので、どちらにも前置きも敬称も無いときは比べません。会社の名前は、語彙表の会社の種類の語（株式会社、(株)、有限会社）の隣にあるものを、品詞と関係なく比べます。名前が同じで種類の語が違うもの（株式会社 と 有限会社）、種類の語を名前の反対側に置いたもの（株式会社アオバ と アオバ株式会社）、記号・幅だけが違うものを指します。同じ種類の語の略した形（株式会社 と (株)）、種類の語を省いた呼び名（アオバシステム、以下「アオバ」）は指しません。名前の一字違いは、7 字以上の名前で、違う字が漢字でないときに限ります（日本電気 と 日本電機 は別の会社のことがある）。漢字だけの名前は、品詞解析が固有名詞と読んだときだけ会社の名前と見ます（清算株式会社、当該株式会社）。場所の名前は、語彙表の場所の語（口、駅、センター、ホール）で終わる名前を、品詞と関係なく比べます。記号・幅だけが違うもの（コンベンションセンター と コンベンション・センター）を指し、名前の読みが同じかかなで書いた読みのもの（筑紫口 と ちくし口）と、固有名詞の字の一字違い（八重洲中央口 と 八重州中央口）は、多いほうが二度以上、少ないほうが一度だけのときに限ります。方角や数字の違い（北口 と 南口、第1会議室 と 第2会議室）は別の所なので指しません。表の升は升ごとに品詞解析し、名前の形をした升（空白を挟んだ二つか三つの漢字・カタカナの語）を本文の名前と比べます。記号・幅だけの違い（佐々木 美穂 と 佐々木　美穂）と、本文と同じく読みが同じで一語だけ違うもの（佐々木 美穂 と 佐々木 美保）を指し、後者は本文の形が二度以上、升の形が一度だけのときに限ります。読みの違う一字違い（佐藤 太郎 と 佐藤 次郎）は別の人のことが多いので指しません（英字の名前は一字違いも指します）。場所の名前は表の升の中のものも比べます。英字の人の名前は、名が同じで、7 字以上の姓の頭 4 字以上がそろい、あいだの 2 字（長いほうは 3 字）までだけ違うもの（Rachel Whitford と Rachel Whitfield）も、多いほうが二度以上、少ないほうが一度だけのときに指します。語の順を入れ替えた名前（統計学基礎 と 基礎統計学）は、同じ語（形態素）を違う順に並べ、どちらも他方を含まず、どちらかが括弧の中か「前提科目：」のような欄の後ろか表の升まるごとに書いた名前のときだけ指します（統計学基礎 と 統計学応用 は別の名前）。
-  en: Only words the tagger reads as proper nouns are checked. Names in capitals only (ACME INC, NASA) are left out. Names one letter apart are reported only when six letters or longer, with the usual form used at least twice and the other once (Iran and Iraq are different names). Two Japanese person names are compared only when written with another form of the same character (斎藤 and 斉藤), or read the same with one character different, the usual form used at least twice and the other once (松本 and 松元); 伊藤 and 伊東 can be two people. A person's name is one the tagger reads as such, or one followed by 様, さん or 氏 (when a name holds a character form the tagger cannot read, the two to four kanji before the honorific are taken as the name). A name without an honorific is compared too when it is two to four kanji standing alone, one form has a lead such as 担当の or 私、 or an honorific, and the other stands before a word such as です, が or までご連絡 (担当の斎藤です and 斉藤までご連絡ください). Places have such pairs as well (鹿島 and 鹿嶋), so two forms with neither a lead nor an honorific are not compared. A company name is one next to a company form from the lexicon (Inc., Ltd., Co., Ltd., LLC), whatever the tagger reads it as. The same name with forms of two kinds (Inc. and Ltd.), and the same name written with other punctuation or width, are reported. The short and long spelling of one form (Inc and Incorporated) and a name written without its form (Aoba Systems, the "Company") are not. Two names one letter apart are reported only when seven letters or longer, with the same form, and the letter is not a kanji. A place name is one ending in a place word from the lexicon (Street, St., Avenue, Station, Center), whatever the tagger reads it as. The same place written with other punctuation (King's Cross and Kings Cross) or with the place word short and long (11th Street and 11th St.) is reported; a place whose name is one letter away (Union Station and Unoin Station) only when the usual form is used at least twice and the other once. Other numbers and other names (1st Street and 2nd Street, North Hall and South Hall) are different places. Table cells are tagged cell by cell, and a cell holding only a name (two to four capitalised words, or two or three kanji or katakana words with a space) is compared with the names read in the prose. Reported are the same name with other punctuation or case, or one word different with the other words the same, either a letter away (Sofia Mendes in the text, Sofia Mendez in the table) or, for a Japanese name, read the same as the prose name (佐々木 美穂 and 佐々木 美保); these only when the prose form is used at least twice and the cell form once. Place names inside table cells are compared too. A word next to a name that the tagger did not read as a name (to Rachel Whitford, with Rachel read as a verb) is read with it when the two together are read as a name elsewhere. Two names with the same given name and a surname of seven letters or more, whose first four letters match and which differ in at most two letters between the shared start and end (three in the longer one), are reported when the usual form is used at least twice and the other once (Rachel Whitford and Rachel Whitfield). A different given name (Daniel Whitfield) or a surname alone is not compared this way. A name with its words in another order (Foundations of Statistics and Statistical Foundations) is reported only when both hold the same words (by stem) and neither contains the other, and one of them is written as a name, in quotation marks, after a label such as "Prerequisite" and a colon, or as a whole table cell. Two Title Case runs alone are not compared (Florida Supreme Court and Supreme Court of Florida), nor are two names joined by "and".
+  ja: 品詞解析が固有名詞と読んだ語だけを見ます。大文字だけの名前（ACME INC、NASA）は見ません。一字違いの英字の名前は、6 字以上で、多いほうが二度以上、少ないほうが一度だけのときに限ります（Iran と Iraq は別の名前）。人の名前どうしは、字体の違う同じ字（斎藤 と 斉藤、渡辺 と 渡邊）か、読みが同じで一字だけ違い、多いほうが二度以上、少ないほうが一度だけのとき（松本 と 松元）に限ります（伊藤 と 伊東 は別の人のことがある）。解析器が読みを持たない名前の字（健汰 の 汰）は、語彙表 name-char-reading の読みで読みます（健太 と 健汰。表の升の名前も同じ）。人の名前とは、品詞解析が人名と読んだ語か、「様」「さん」「氏」の付いた名前です（解析器が名前と読めない字体の字を含む名前は、敬称の前の 2〜4 字の漢字を名前と読みます）。敬称は名前とのあいだの空白を越えて見ます。品詞解析が名と読まない語も、姓のあとに空白を挟んで書き、敬称が付けば名と読みます（森下 千尋 様）。敬称の無い名前も、前後に漢字が続かない 2〜4 字の漢字で、片方の書き方に「担当の」「私、」などの前置きか敬称が付き、もう片方が「です」「が」「までご連絡」などの前にあれば比べます（担当の斎藤です と 斉藤までご連絡ください）。地名にも同じ組がある（鹿島 と 鹿嶋）ので、どちらにも前置きも敬称も無いときは比べません。会社の名前は、語彙表の会社の種類の語（株式会社、(株)、有限会社）の隣にあるものを、品詞と関係なく比べます。名前が同じで種類の語が違うもの（株式会社 と 有限会社）、種類の語を名前の反対側に置いたもの（株式会社アオバ と アオバ株式会社）、記号・幅だけが違うものを指します。同じ種類の語の略した形（株式会社 と (株)）、種類の語を省いた呼び名（アオバシステム、以下「アオバ」）は指しません。名前の一字違いは、7 字以上の名前で、違う字が漢字でないときに限ります（日本電気 と 日本電機 は別の会社のことがある）。漢字を含む名前は、7 字に満たなくても、4 字以上のかなの語の中のかな一字違いを、多いほうが二度以上、少ないほうが一度だけのときに指します（コモレビ電機 と コモレピ電機）。名前の漢字一字を、品詞解析が言うその読みのかなで書いた所（みどり野生命 と みどりの生命）も、多いほうが二度以上、かなのほうが一度だけのときに指します。読みは、かなと漢字一字の語からだけ取ります。漢字だけの名前は、品詞解析が固有名詞と読んだときだけ会社の名前と見ます（清算株式会社、当該株式会社）。場所の名前は、語彙表の場所の語（口、駅、センター、ホール）で終わる名前を、品詞と関係なく比べます。記号・幅だけが違うもの（コンベンションセンター と コンベンション・センター）を指し、名前の読みが同じかかなで書いた読みのもの（筑紫口 と ちくし口）と、固有名詞の字の一字違い（八重洲中央口 と 八重州中央口）は、多いほうが二度以上、少ないほうが一度だけのときに限ります。方角や数字の違い（北口 と 南口、第1会議室 と 第2会議室）は別の所なので指しません。表の升は升ごとに品詞解析し、名前の形をした升（空白を挟んだ二つか三つの漢字・カタカナの語）を本文の名前と比べます。記号・幅だけの違い（佐々木 美穂 と 佐々木　美穂）と、本文と同じく読みが同じで一語だけ違うもの（佐々木 美穂 と 佐々木 美保）を指し、後者は本文の形が二度以上、升の形が一度だけのときに限ります。読みの違う一字違い（佐藤 太郎 と 佐藤 次郎）は別の人のことが多いので指しません（英字の名前は一字違いも指します）。場所の名前は表の升の中のものも比べます。英字の人の名前は、名が同じで、7 字以上の姓の頭 4 字以上がそろい、あいだの 2 字（長いほうは 3 字）までだけ違うもの（Rachel Whitford と Rachel Whitfield）も、多いほうが二度以上、少ないほうが一度だけのときに指します。語の順を入れ替えた名前（統計学基礎 と 基礎統計学）は、同じ語（形態素）を違う順に並べ、どちらも他方を含まず、どちらかが括弧の中か「前提科目：」のような欄の後ろか表の升まるごとに書いた名前のときだけ指します（統計学基礎 と 統計学応用 は別の名前）。製品の名前は、語彙表の製品の形の語（錠、カプセル、クリーム）のすぐ前の 4 字以上のカタカナを、品詞と関係なく比べます（見出しと表の升の中のものも数えます）。形の語が同じで、名前が一字の置き換えか長音の記号一つの有る無しだけ違い（ミナモール錠 と ミナモル錠）、多いほうが二度以上、少ないほうが一度だけのときに指します。名前が同じで形の語だけが違うもの（ミナモール錠 と ミナモール顆粒）は、同じ製品の別の形のことが多いので指しません。一語の名前の中で同じ音を書く字（桜ヶ丘・桜ケ丘、霞ヶ関・霞が関）は、漢字に挟まれたときだけ同じ字と見ます。品詞解析が一語と読まない書き方（桜・が・丘）は見ません。字を抜いた形（桜丘）や、違う漢字の名前（桜ヶ岡）は別の名前です。この違いだけは、表の升の中の名前とも比べます。型番（英大文字と数字を合わせた 4 字以上の語）は、文書の題か、語彙表の型番の欄の語（型番、品番、型式）の後ろに書いたものだけを文書の型番と読み、文書の中に二度以上出てくるとき、ハイフン・空白・字の大小・幅だけを変えた形（KM-SP300 と KM-SP-300、KMSP300、km-sp300）を指します。字や数の違う型番（KM-SP300 と KM-SP310、比べる表の兄弟機種、部品表の品番）は別の型番なので指しません。英小文字を含む語（iOS17）、版（v1.2 と v1.20）、より長い型番・URL・パス・ファイル名の一部（KM-SP300-B、km-sp300.pdf）は型番と読みません。空白で区切って書く規格の番号（JIS C 8714）は一つの語でないので、題や欄に書いても型番と読みません。一度ずつ書いた二つの形は、片方だけが人を書く欄（語彙表 person-label の語と「：」の後ろで、名前で行が終わるもの。氏名：山下 美咲）にあれば、欄の形を正しい形と見て、もう片方を同じ決まりで比べます（山下 美咲 と 山下 未咲）。どちらも欄にあるか、どちらも欄に無ければ比べません。欄に書いた名と同じ名を空白の後ろに書いたもの（受託者：西村 亮 の文書の 西邑 亮が）は、敬称が無くても人の名前と読みます。読みの違う一字違いでも、姓と名のそろった名前で、違う二字が語彙表 name-lookalike-char の形の似た字の組（太 と 大、未 と 末）なら、多いほうが二度以上、少ないほうが一度だけのときに指します（山本 健太 と 山本 健大。名だけの名前は兄弟のことがあるので比べません）。
+  en: Only words the tagger reads as proper nouns are checked. Names in capitals only (ACME INC, NASA) are left out. Names one letter apart are reported only when six letters or longer, with the usual form used at least twice and the other once (Iran and Iraq are different names). Two Japanese person names are compared only when written with another form of the same character (斎藤 and 斉藤), or read the same with one character different, the usual form used at least twice and the other once (松本 and 松元); 伊藤 and 伊東 can be two people. A name character the tagger has no reading for (汰 in 健汰) is read with its reading from the lexicon name-char-reading (健太 and 健汰), in table cells as well. A person's name is one the tagger reads as such, or one followed by 様, さん or 氏 (when a name holds a character form the tagger cannot read, the two to four kanji before the honorific are taken as the name). The honorific may follow the name after a space. A word the tagger does not read as a given name is taken as one when it follows a surname after a space and carries an honorific (森下 千尋 様). A name without an honorific is compared too when it is two to four kanji standing alone, one form has a lead such as 担当の or 私、 or an honorific, and the other stands before a word such as です, が or までご連絡 (担当の斎藤です and 斉藤までご連絡ください). Places have such pairs as well (鹿島 and 鹿嶋), so two forms with neither a lead nor an honorific are not compared. A company name is one next to a company form from the lexicon (Inc., Ltd., Co., Ltd., LLC), whatever the tagger reads it as. The same name with forms of two kinds (Inc. and Ltd.), and the same name written with other punctuation or width, are reported. The short and long spelling of one form (Inc and Incorporated) and a name written without its form (Aoba Systems, the "Company") are not. Two names one letter apart are reported only when seven letters or longer, with the same form, and the letter is not a kanji. A name holding kanji is compared when shorter too, for one kana apart inside a kana word of four or more (コモレビ電機 and コモレピ電機), with the usual form used at least twice and the other once. One kanji of a name written in its kana reading (みどり野生命 and みどりの生命) is reported the same way; the reading is the tagger's, taken only from a word of kana and one kanji. A place name is one ending in a place word from the lexicon (Street, St., Avenue, Station, Center), whatever the tagger reads it as. The same place written with other punctuation (King's Cross and Kings Cross) or with the place word short and long (11th Street and 11th St.) is reported; a place whose name is one letter away (Union Station and Unoin Station) only when the usual form is used at least twice and the other once. Other numbers and other names (1st Street and 2nd Street, North Hall and South Hall) are different places. Table cells are tagged cell by cell, and a cell holding only a name (two to four capitalised words, or two or three kanji or katakana words with a space) is compared with the names read in the prose. Reported are the same name with other punctuation or case, or one word different with the other words the same, either a letter away (Sofia Mendes in the text, Sofia Mendez in the table) or, for a Japanese name, read the same as the prose name (佐々木 美穂 and 佐々木 美保); these only when the prose form is used at least twice and the cell form once. Place names inside table cells are compared too. A word next to a name that the tagger did not read as a name (to Rachel Whitford, with Rachel read as a verb) is read with it when the two together are read as a name elsewhere. Two names with the same given name and a surname of seven letters or more, whose first four letters match and which differ in at most two letters between the shared start and end (three in the longer one), are reported when the usual form is used at least twice and the other once (Rachel Whitford and Rachel Whitfield). A different given name (Daniel Whitfield) or a surname alone is not compared this way. A name with its words in another order (Foundations of Statistics and Statistical Foundations) is reported only when both hold the same words (by stem) and neither contains the other, and one of them is written as a name, in quotation marks, after a label such as "Prerequisite" and a colon, or as a whole table cell. Two Title Case runs alone are not compared (Florida Supreme Court and Supreme Court of Florida), nor are two names joined by "and". In Japanese, a product name is four or more katakana right before a product-form word from the lexicon (錠, カプセル, クリーム), whatever the tagger reads it as, counted in headings and table cells as well. Two such names with the same form word, differing by one replaced kana or one long-vowel mark (ミナモール錠 and ミナモル錠), are reported when the usual form is used at least twice and the other once. The same name with another form word (ミナモール錠 and ミナモール顆粒) is not, as a product line often has both. In a Japanese name, a character for the same sound (桜ヶ丘 and 桜ケ丘, 霞ヶ関 and 霞が関) is the same only between two kanji, in a name the tagger reads as one word (not 桜が丘, read as 桜, が, 丘). A name without it (桜丘) or with other kanji (桜ヶ岡) is another name. This difference alone is also looked for in the names inside table cells. A model number (four or more capitals and digits, as one word) is read as the document's model only when it is in the title or after a model label from the lexicon (Model, Model No., Part No.); when it appears twice or more, a form that changes only hyphens, spaces, case or width (KM-SP300 and KM-SP-300, KMSP300, km-sp300) is reported. Codes with other letters or digits (KM-SP300 and KM-SP310, a sibling model in a comparison table, the part numbers of a parts list) are different models and are not compared. A word with small letters (iOS17), a version (v1.2 and v1.20), and a code inside a longer code, a URL, a path or a file name (KM-SP300-B, km-sp300.pdf) are not read as model numbers. A standard number written with spaces (JIS C 8714) is not one word, so it is not read as a model even in the title or after a label. Two forms written once each are compared when exactly one of them is in a field that names a person (a label from the lexicon person-label and a colon, with the name ending the line, as in "Contractor:" followed by Priya Raman); the field's form is taken as the usual one and the other is held to the same rules (Priya Raman and Priya Rahman). Two forms both in such fields, or neither, are not compared. Read differently, two full names (surname and given name) are still reported when they differ by one character and the two characters are a lookalike pair from the lexicon name-lookalike-char (太 and 大, 未 and 末), with the usual form used at least twice and the other once (山本 健太 and 山本 健大); a given name alone is not compared this way, as siblings can have such names.
 rewrite:
   depth: light
   ja:

package/rules/net-amount-mismatch.yaml

@@ -0,0 +1,57 @@
+id: net-amount-mismatch
+layer: L3
+status: stable
+how_to_find: net-amount-mismatch
+# The names are lexicon net-amount-label (groups gross, deduction, deduction-part, net, taxable); terms cut off a name are net-amount-period; words of magnitude are amount-multiplier.
+severity: warning
+
+name:
+  ja: 差引額が、総額から控除合計を引いた額と合わない
+  en: A net amount that is not the gross amount minus the deductions
+
+why:
+  ja: 給与明細の「差引支給額」は、総支給額から控除合計を引いた額です。受け取る人は差引額を見て、引き算し直しません。控除を直したのに差引額を直さなかったとき、前の月の明細から写したとき、数字を打ち間違えたときに食い違います。同じ表か同じ節に、差し引く前の額（総支給額）、控除の合計（控除合計）、差し引いた後の額（差引支給額）の名前が一つずつあり、それぞれに金額があれば、差引額を引き算と比べます。表では、見出しの行に三つの名前が並ぶ表は行ごとに、一列目に三つの名前が並ぶ表は列ごとに比べます。表の外では、節（見出しから次の見出しまで）の「総支給額：117,380円」のような名前と金額だけの行を読みます。源泉徴収の明細のように差引額の名前が無く、課税対象額（課税所得）がある表や節では、課税対象額を同じ引き算と比べます。差引額と課税対象額が両方あるときは、控除合計が支払いから引く額なので、差引額だけを比べます。不動産の売買代金の精算書（売買代金 − 手付金 − 融資金 = 残代金）のように、控除の合計の代わりに差し引く額が一つずつ書いてあれば、それらを足して引きます。合計と内訳が両方あれば、合計だけを引きます。どの語が三つのどれかは、言語パッケージの語彙表 net-amount-label にあります。通貨の印が付いた金額は一円（一セント）まで合わなければ言います。千円や百万円の単位の金額、印の無い数は丸めてあるかもしれないので、最後の桁の一つ分までのずれは言いません。
+  en: >-
+    The "Net pay" of a payslip is the gross pay minus the total deductions. The person paid reads the net and does not subtract again. It goes wrong when a deduction is corrected and the net is not, when the net is copied from last month's payslip, or when a digit is mistyped. When one table or one section names the amount before deductions (Gross pay), the total deductions (Total deductions) and the amount after them (Net pay) once each, each with an amount, the net is compared with the subtraction. A table whose header row has the three names is compared row by row, and one whose first column has them column by column. Outside tables, the lines of a section (from one heading to the next) that are only a name and an amount ("Gross pay: $1,661.00") are read. Where no net is named but a taxable amount is (Taxable income, as on a tax statement), the taxable amount is compared with the same subtraction. Where both are named, the deductions are what is paid out, so only the net is compared. Where the deductions are written one by one instead of as a total, as on a property settlement statement (Purchase price − Earnest money deposit − Loan proceeds = Balance due), they are added up; where a total is written beside them, only the total is taken. Which words name the three is in the language package's lexicon net-amount-label. Amounts with a currency mark must agree to the cent. Amounts in thousands or millions, and numbers without a mark, may be rounded, so a difference of one step of their last digit is not reported.
+
+message:
+  ja: 「{written}」は、総額から控除合計を引くと {expected} です
+  en: '"{written}" is written, but the gross amount minus the deductions is {expected}'
+
+how_to_fix:
+  ja: 総額、控除合計、差引額のどれが正しいかを確かめて、ほかを直してください。
+  en: Check which of the gross amount, the deductions and the net is right, and correct the others.
+
+levels: { strict: error, normal: warning, relaxed: info }
+
+use_for: [business, technical, blog, legal, docs, academic, literature, speech]
+off_for:
+  literature: "Length, rhythm, repetition and register are the author's choices in literature, and dialogue mixes styles. Only what one work can hold itself to runs: consistent spacing and commas, and numbering without gaps."
+
+group: facts
+summary:
+  ja: 差引支給額が、総支給額 − 控除合計 と合わない所
+  en: A net pay that is not the gross pay minus the total deductions
+example:
+  ja:
+    before: |
+      - 総支給額：117,380円
+      - 控除合計：3,074円
+      - 差引支給額：114,036円
+    after: |
+      - 総支給額：117,380円
+      - 控除合計：3,074円
+      - 差引支給額：114,306円
+  en:
+    before: |
+      - Gross pay: $1,661.00
+      - Total deductions: $264.59
+      - Net pay: $1,369.41
+    after: |
+      - Gross pay: $1,661.00
+      - Total deductions: $264.59
+      - Net pay: $1,396.41
+
+not_flagged:
+  ja: 三つの名前のどれかが無い表や節、同じ名前に違う金額が二つある表の列や節、名前が別々の表や節にあるもの（支給の表の総支給額と控除の表の控除合計）、三つの単位が違うもの（円とドル、千円と円）、数でない金額（—、▲、括弧の付いた値）は比べません。文の中の金額（「差引支給額は114,306円です」）は読みません。語彙表に無い名前（振込額、現金支給額）は読みません。総額より控除が大きいときも言いません。
+  en: 'A table or section without one of the three names, a table column or section with two different amounts under one name, names spread over separate tables or sections (the gross in the earnings table and the total in the deductions table), amounts in different units (yen and dollars, thousands and units), and a value that is not a plain number (—, a negative or bracketed value) are not compared. An amount inside a sentence ("Net pay is $1,396.41") is not read. A name not in the lexicon (Amount deposited, Cash paid) is not read. Deductions larger than the gross amount are not reported either.'

package/rules/net-over-gross.yaml

@@ -0,0 +1,57 @@
+id: net-over-gross
+layer: L3
+status: stable
+how_to_find: net-over-gross
+# The net and gross columns are found by the language's net-gross-weight-column lexicon; g, kg and t are converted by unit-mass.
+extra_word_lists: [net-gross-weight-column, unit-mass]
+severity: warning
+
+name:
+  ja: 正味重量が総重量より重い
+  en: A net weight larger than the gross weight
+
+why:
+  ja: 梱包明細書（パッキングリスト）の正味重量は品物だけの重さで、総重量は箱や梱包材を含めた重さです。同じ行で正味重量が総重量を上回ることはありません。正味重量を写し間違えたとき、別の箱の行と取り違えたとき、総重量だけを直したときに起きます。どの列が正味重量で、どの列が総重量かは、列の見出しの語（正味重量、純重量、N.W.、総重量、グロス重量、G.W.）だけで決めます。列の並びでは決めません。比べるのは、どちらの升も書いた数（96.0kg）である行で、合計の行も比べます。g・kg・t は語彙表の単位で換算して比べ、語彙表に無い単位は同じ単位どうしだけを比べます。
+  en: "On a packing list, the net weight is the goods alone and the gross weight includes the cartons and packing, so the net weight of a row cannot exceed its gross weight. It does when the net weight is copied wrong, when a row is mixed up with another carton, or when only the gross weight is corrected. Which column is net and which is gross is decided by the heading words alone (Net weight, N.W., Gross weight, G.W.), never by where a column stands. Only rows where both cells are written numbers (96.0 kg) are compared, the totals row too. Grams, kilograms, tonnes and pounds are converted by the units word list; other units are compared only with the same unit."
+
+message:
+  ja: 正味重量「{net}」が、同じ行の総重量「{gross}」より重くなっています
+  en: 'The net weight "{net}" is larger than the gross weight "{gross}" on the same row'
+
+how_to_fix:
+  ja: 正味重量と総重量のどちらが正しいかを確かめて、もう一方を直してください。
+  en: Check which of the net and gross weights is right and correct the other.
+
+levels: { strict: error, normal: warning, relaxed: info }
+
+use_for: [business, technical, blog, legal, docs, academic, literature, speech]
+off_for:
+  literature: "Length, rhythm, repetition and register are the author's choices in literature, and dialogue mixes styles. Only what one work can hold itself to runs: consistent spacing and commas, and numbering without gaps."
+
+group: facts
+summary:
+  ja: 表の同じ行で、正味重量が総重量より重い所
+  en: A row of a packing list whose net weight is larger than its gross weight
+example:
+  ja:
+    before: |
+      | 箱番号 | 品名 | 正味重量 | 総重量 |
+      | --- | --- | --- | --- |
+      | No.1 | 軸受けハウジング | 106.0kg | 104.0kg |
+    after: |
+      | 箱番号 | 品名 | 正味重量 | 総重量 |
+      | --- | --- | --- | --- |
+      | No.1 | 軸受けハウジング | 96.0kg | 104.0kg |
+  en:
+    before: |
+      | Carton | Description | Net weight | Gross weight |
+      | --- | --- | --- | --- |
+      | No. 1 | Control unit | 68.0 kg | 65.0 kg |
+    after: |
+      | Carton | Description | Net weight | Gross weight |
+      | --- | --- | --- | --- |
+      | No. 1 | Control unit | 58.0 kg | 65.0 kg |
+
+not_flagged:
+  ja: 正味重量の列か総重量の列が無い表、どちらかの列が二つ以上ある表、正味と総重量の両方を言う見出しの列がある表は比べません。二つの見出しが正味と総重量のほかに違うことを言う表（1個あたりの正味重量と1箱あたりの総重量）も比べません。升が空、「-」、「未定」、符号やかっこ付き、「約」「以上」「程度」の付いた行、二つの升の単位を換算できない行も比べません。正味重量が総重量と同じか軽い行は言いません。
+  en: "A table with no net or no gross column, or with two of either, is not compared. Neither is a table with a heading that names both net and gross, or whose two headings say something different besides net and gross (net weight per piece beside gross weight per carton). A row whose cell is blank, a dash, TBD, signed, bracketed, approximate or a bound (100 kg or more), or whose two units cannot be converted to each other, is skipped. A net weight equal to or less than the gross weight is not reported."

package/rules/no-doubled-joshi.yaml

@@ -35,7 +35,7 @@ sources: [yamamoto-cheng-2015]
 group: readability
 summary:
   ja: 「会社の部の課の方針」のように、同じ助詞が区切りなしに続く所
-  en: The same Japanese particle repeated with no break, so the reader cannot tell what attaches to what
+  en: The same Japanese particle repeated, so what attaches to what is unclear
 example:
   ja:
     before: 本社の営業部の第二課の方針を確認しました。

package/rules/no-em-dash.yaml

@@ -13,8 +13,8 @@ name:
   en: Too many dashes
 
 why:
-  ja: 日本語の組版でダッシュは扱いが難しく、環境によって長さも見えかたも変わります。読点や括弧で書けることが多いです。
-  en: In English a dash is a legitimate tool, and a heavy hand with it is also one of the better-known marks of generated text. On its own it says little; it counts toward the composite signal.
+  ja: 日本語の組版でダッシュは扱いが難しく、環境によって長さも見えかたも変わります。読点や括弧で書けることが多いです。数字のあいだで範囲を示すダッシュ（45–58、10:00–10:15）は数えません。
+  en: In English a dash is a legitimate tool, and a heavy hand with it is also one of the better-known marks of generated text. On its own it says little; it counts toward the composite signal. An en dash between two numbers (45–58, 10:00–10:15) marks a range and is not counted.
 
 message:
   ja: 1000 字あたり {density} 個のダッシュがあります（{limit} 個まで）

package/rules/no-mixed-desumasu.yaml

@@ -42,7 +42,7 @@ sources: [yasashii-nihongo-2020, koyobun-2022, jtf-style-guide]
 group: consistency
 summary:
   ja: 「です・ます」の中に「だ・である」が混ざった文
-  en: A Japanese sentence whose ending (polite or plain) differs from the rest of the document
+  en: A Japanese sentence whose polite or plain ending differs from the rest
 example:
   ja:
     before: |

package/rules/no-nakaguro-parallel.yaml

@@ -32,7 +32,7 @@ off_for:
 group: readability
 summary:
   ja: 一つの文に「・」で並べた語が多すぎて、どれとどれが組なのか分からない所
-  en: So many items joined by middle dots (・) in one Japanese sentence that the groups blur
+  en: So many items joined by middle dots (・) in one sentence that groups blur
 example:
   ja:
     before: 対象は営業・企画・総務・経理・人事・法務の各部と、東京・大阪・名古屋の各拠点です。

package/rules/nominalization.yaml

@@ -42,7 +42,7 @@ sources: [yasashii-nihongo-2020, plain-2011, gov-uk-clear-language, reinhart-202
 group: wording
 summary:
   ja: 「調査を実施した」「確認を行う」のように、動詞を名詞にして別の動詞に運ばせた言い方が、文書の長さに対して多い所
-  en: Verbs hidden in nouns ("make a decision", "conduct an analysis"), many for the document's length
+  en: Verbs hidden in nouns ("make a decision", "conduct an analysis"), used too often
 example:
   ja:
     pad: true

package/rules/nonstandard-word.yaml

@@ -41,7 +41,7 @@ off_for:
 group: wording
 summary:
   ja: 「違かった」「すいません」「irregardless」のように、書き言葉では標準的でない語
-  en: A word that formal writing treats as nonstandard, such as "irregardless" or "comprised of"
+  en: A word formal writing treats as nonstandard, such as "irregardless"
 example:
   ja:
     before: 昨日の数字は、先週の報告と違かった。

package/rules/number-style-consistency.yaml

@@ -50,7 +50,7 @@ off_for:
 group: consistency
 summary:
   ja: 3件と三件、1,000 と 1000、10% と 10％ のように、一つの文書で数の書き方が混ざっている所
-  en: One document writing numbers two ways, such as 1,000 and 1000, 10% and 10 percent, or three files and 3 files
+  en: Numbers written two ways in one document, such as 1,000 and 1000
 
 example:
   ja:
@@ -67,8 +67,8 @@ example:
       We fixed three bugs, closed four issues and merged two pull requests. The release also adds five tests.
 
 not_flagged:
-  ja: 鉤括弧や引用符の中の数、注や URL を含む文、日本語の中の英文は数えません。「一つ」「一人」のような一だけの漢数字と、「万人」「千人」のような位の字だけの漢数字、「第三回」「統一」のように漢字の後ろの数、日付の日・月・年は数えません。位取りは、通貨の記号か数え方の語・単位が付いた大きな数だけを見て、年に見える 4 桁の数（2019 年）、番号（H.R. 5376）、版や電話番号は見ません。全角と半角の違いは fullwidth-alnum-consistency が見ます。法務・文学・話し言葉のジャンルでは見ません。
-  en: 'Numbers inside quotation marks, sentences with a note mark or a URL, and text in another language are not counted. A number word that opens a sentence (spelt out by every style), "one", and a number before a unit of measure (6 months, 3 miles) are not counts. Grouping is checked only for large numbers with a currency sign or a word after them; a four-digit number that looks like a year, a label number (H.R. 5376, Form 1040), versions and phone numbers are left out. Legal, literary and speech genres do not run this rule.'
+  ja: 鉤括弧や引用符の中の数、注や URL を含む文、日本語の中の英文は数えません。「一つ」「一人」のような一だけの漢数字と、「万人」「千人」のような位の字だけの漢数字、「第三回」「統一」のように漢字の後ろの数、日付の日・月・年、英語で名前の一部になった数（Grade 5 students、Room 12）は数えません。位取りは、通貨の記号か数え方の語・単位が付いた大きな数だけを見て、年に見える 4 桁の数（2019 年）、番号（H.R. 5376）、版や電話番号は見ません。全角と半角の違いは fullwidth-alnum-consistency が見ます。法務・文学・話し言葉のジャンルでは見ません。
+  en: 'Numbers inside quotation marks, sentences with a note mark or a URL, and text in another language are not counted. A number word that opens a sentence (spelt out by every style), "one", a number before a unit of measure (6 months, 3 miles) and a number that names a group (Grade 5 students, Room 12) are not counts. Grouping is checked only for large numbers with a currency sign or a word after them; a four-digit number that looks like a year, a label number (H.R. 5376, Form 1040), versions and phone numbers are left out. Legal, literary and speech genres do not run this rule.'
 level_meaning:
   ja: 少ないほうの書き方が、その種類の数の {limit}% 以下のとき指摘
   en: reported when the less common style is {limit}% of its kind or less

package/rules/number-without-unit.yaml

@@ -27,7 +27,7 @@ use_for: [business, technical, blog, legal, docs, academic, literature, speech]
 group: structure
 summary:
   ja: 同じ列や同じ見出しの語のほかの値には単位があるのに、数だけを書いた値
-  en: A bare number where the other values of the same column or label all carry one unit
+  en: A bare number where the other values in its column all have a unit
 example:
   ja:
     before: |

package/rules/number-word-figure-mismatch.yaml

@@ -29,7 +29,7 @@ use_for: [business, technical, blog, legal, docs, academic, literature, speech]
 group: facts
 summary:
   ja: 「三（4）か月」「金参拾万円（30,000円）」のように、文字で書いた数と後ろの括弧の数字が違う所
-  en: A number written in words followed by a figure in brackets that says another number, such as "six (7) months"
+  en: A number in words with a different figure in brackets, as "six (7) months"
 example:
   ja:
     config: { genre: legal/contract }

package/rules/numbering-gap.yaml

@@ -28,7 +28,7 @@ use_for: [business, technical, blog, legal, docs, academic, literature, speech]
 group: structure
 summary:
   ja: 第1条の次が第3条、(a) の次が (c) のように、番号が飛んだり重なったりしている所
-  en: A numbered item that skips or repeats a number, such as Article 3 right after Article 1
+  en: A numbered item that skips or repeats a number, such as Article 1 then 3
 example:
   ja:
     before: |

package/rules/obligation-without-deadline.yaml

@@ -35,7 +35,7 @@ off_for:
 group: wording
 summary:
   ja: ほかの義務に期限を書いた契約で、支払う・引き渡す・返還する・通知する義務に期限を書いていない所
-  en: A duty to pay, deliver, return or notify with no deadline, in a contract that gives its other duties one
+  en: A duty to pay, deliver, return or notify with no deadline, where others have one
 example:
   ja:
     config: { genre: legal/contract }

package/rules/ordinal-suffix-mismatch.yaml

@@ -27,7 +27,7 @@ use_for: [business, technical, blog, legal, docs, academic, literature, speech]
 group: wording
 summary:
   ja: 「3th」「22th」「11st」のように、英語の序数の字が数と合わない所
-  en: An English ordinal whose ending does not go with its number, such as "3th", "22th" or "11st"
+  en: An ordinal whose ending does not fit its number, such as "3th" or "11st"
 example:
   ja:
     before: 今年で 3th イベントになります。

package/rules/orthographic-variant.yaml

@@ -41,8 +41,8 @@ off_for:
 
 group: consistency
 summary:
-  ja: 「出来る」と「できる」、「引っ越し」と「引越し」のように、同じ語を一つの文書の中で二通りに書いている所（語の一覧なしで見つけます）
-  en: One word written two ways in one document (e-mail and email, GitHub and Github), found without a word list
+  ja: 「出来る」と「できる」のように、同じ語を一つの文書の中で二通りに書いている所
+  en: One word written two ways in one document, such as e-mail and email
 example:
   ja:
     before: |

package/rules/outside-announced-list.yaml

@@ -39,7 +39,7 @@ use_for: [business, technical, blog, legal, docs, academic, literature, speech]
 group: consistency
 summary:
   ja: 「対応 OS は Windows と macOS」と挙げたあとで、一覧に無い「Linux」に対応すると書いている所
-  en: A later sentence naming a platform or language that the list given earlier does not have
+  en: A platform or language named later that the list given earlier left out
 example:
   ja:
     before: |

package/rules/padded-intro.yaml

@@ -40,7 +40,7 @@ off_for:
 group: wording
 summary:
   ja: 「近年、〜が注目されています」のように、どの文書にも当てはまる書き出し
-  en: An opening such as "In recent years, X has gained attention" that fits any document
+  en: An opening such as "In recent years, X has gained attention" that says nothing
 example:
   ja:
     before: 近年、生成 AI が注目を集めています。今回、社内の問い合わせ対応に試しました。

package/rules/paragraph-restatement.yaml

@@ -33,7 +33,7 @@ off_for:
 group: readability
 summary:
   ja: 「つまり」「このように」で始まり、前の段落の語をほとんど繰り返すだけの段落
-  en: A paragraph that opens with "In other words" or "In short" and mostly repeats the paragraph before
+  en: A paragraph opening "In other words" that mostly repeats the one before
 
 example:
   ja:

package/rules/party-role-name.yaml

@@ -34,7 +34,7 @@ off_for:
 group: consistency
 summary:
   ja: 当事者を「甲」「乙」と定めたあとで、「委託者」「受託者」のような立場の語で書いている所
-  en: A party given a short name ("the Supplier") and later called by a role word (the Vendor)
+  en: A party given a short name ("the Supplier") and later called something else
 example:
   ja:
     config: { genre: legal/contract }
@@ -66,8 +66,8 @@ example:
       1. The Customer pays the Supplier within thirty days of each invoice.
 
 not_flagged:
-  ja: 当事者の呼び名を定める前（「売主 有限会社たちばな木工（以下「甲」という。）」の売主）、定めた語そのもの（「開示者」と定義した文書の開示者）、定義した語の一部、長い語の一部（再委託先、委託者側）は見ません。当事者の定義は、呼び名が甲・乙・当社のような語か立場の語であるもの、または括弧のすぐ前に会社の形（株式会社、Ltd）があるものだけを読みます。法務のジャンルで見ます。法令・規程は当事者を定めず、立場の語で人を呼ぶので見ません。
-  en: 'Uses before the parties are named, a role word the document defines itself ("Recipient" means), a role word inside a defined term (Receiving Party) or a longer word, and a lower-case role word (the vendor, an ordinary noun) are not reported. A party is read from a short name in brackets that is a party word (we) or a role word itself, or that follows a company name (Ltd, Inc.). The legal genres run this rule, except statutes, which name people by their role and define no parties.'
+  ja: 当事者の呼び名を定める前（「売主 有限会社たちばな木工（以下「甲」という。）」の売主）、定めた語そのもの（「開示者」と定義した文書の開示者）、定義した語の一部、長い語の一部（再委託先、委託者側）、英語で文の頭に冠詞なしで書いた複数形（Customers who register …。文頭なので大文字になった普通の名詞）は見ません。当事者の定義は、呼び名が甲・乙・当社のような語か立場の語であるもの、または括弧のすぐ前に会社の形（株式会社、Ltd）があるものだけを読みます。法務のジャンルで見ます。法令・規程は当事者を定めず、立場の語で人を呼ぶので見ません。
+  en: 'Uses before the parties are named, a role word the document defines itself ("Recipient" means), a role word inside a defined term (Receiving Party) or a longer word, a lower-case role word (the vendor, an ordinary noun), and a plural with no article at the head of a sentence (Customers who register …, the ordinary noun capitalised by its place) are not reported. A party is read from a short name in brackets that is a party word (we) or a role word itself, or that follows a company name (Ltd, Inc.). The legal genres run this rule, except statutes, which name people by their role and define no parties.'
 rewrite:
   depth: light
   ja:

package/rules/percent-sum-mismatch.yaml

@@ -56,5 +56,5 @@ example:
       - Europe: 40%
 
 not_flagged:
-  ja: 割合の語の無い百分率（利用率、伸び率）、複数回答の集計、符号の付いた増減（+3%）、合計の行がある表（合計が内訳の和と合わない の対象）、項目が一つだけのものは足しません。一つの文の内訳は、和が 100% から 10 ポイントより離れていれば言いません（大きい項目だけを挙げた文や、別の全体・前の年の割合を並べた文）。
-  en: 'Percentages not named as shares (rates, growth), multiple-answer counts, signed changes (+3%), a table with a total row (left to "A total that is not the sum of its items"), and a single item are not added. A breakdown in one sentence whose sum is more than 10 points from 100% is not reported: it names only the larger parts, or shares of different wholes or of two periods.'
+  ja: 割合の語の無い百分率（利用率、伸び率）、見出しが行ごとの率（還元率、割引率、税率）の列（上の見出しが「内訳」でも）、複数回答の集計、符号の付いた増減（+3%）、合計の行がある表（合計が内訳の和と合わない の対象）、項目が一つだけのものは足しません。一つの文の内訳は、和が 100% から 10 ポイントより離れていれば言いません（大きい項目だけを挙げた文や、別の全体・前の年の割合を並べた文）。
+  en: 'Percentages not named as shares (rates, growth), a column headed by a rate applied to each row (Discount, Tax rate) even under a heading that names a breakdown, multiple-answer counts, signed changes (+3%), a table with a total row (left to "A total that is not the sum of its items"), and a single item are not added. A breakdown in one sentence whose sum is more than 10 points from 100% is not reported: it names only the larger parts, or shares of different wholes or of two periods.'

package/rules/period-label-conflict.yaml

@@ -0,0 +1,68 @@
+id: period-label-conflict
+layer: L3
+status: stable
+how_to_find: period-label-conflict
+# Period names and their kinds are reporting-period-frame; the title's results words reporting-results-word; the period
+# line's words reporting-period-line; the words that excuse a name reporting-comparison and reporting-forecast; the words
+# for the current period reporting-current, and phrases that hold one without naming a period reporting-current-not.
+# Years a name marks by itself (令和8年分, Tax Year 2026) are reporting-year-frame, read without a results word; era
+# years become Western years by calendar-era's first_year; words that hold a name's form without being one (分担金) are
+# reporting-period-not.
+severity: warning
+
+name:
+  ja: 題の決算期間と違う期間
+  en: A period other than the one the title reports
+
+why:
+  ja: 「2027年3月期 第2四半期決算短信」の表の見出しが「当第3四半期累計」だと、読み手はどちらの期間の数字なのかを問い合わせることになります。前の期の文書を写して直し忘れたとき、四半期ごとに使い回す表の見出しを直し忘れたときに起きます。題（最初の行）が「決算」「業績」の語と一緒に書いた期間（第2四半期、2027年3月期、上半期、FY2026）を文書の期間とし、題に無い種類は頭の「対象期間：」の一行から補います。その後ろの今の業績を言う所、つまり表の見出しの升と、「当期」「当第」「業績」のある文と見出しに、同じ種類の別の期間があれば指します。前年同期、前期、前年同四半期、予想、見通し、次期のような比べる相手や先の期間の語がその升、文、節の見出しにあれば比べません。文書の期間と一緒に書いた期間（| 2025年3月期 | 2026年3月期 |）も比べる相手として読みます。源泉徴収の明細書のように、題が「令和8年分」「2026年分」と年を言う文書は、決算の文書でなくても読み、本文のどこかに「2025年分」と別の年分があれば指します。元号の年は西暦にして比べるので、令和8年分と2026年分は同じ年です。
+  en: 'When a release titled "Second Quarter Results" heads its table "Q3 FY2027", the reader has to ask which period the figures are for. It happens when last period''s release is copied and one label is not updated, or when a table reused every quarter keeps its old heading. The period the title (the first line) names together with a results word ("Results", "Earnings") is the document''s period: a quarter, a half or a fiscal year (Q2, first half, FY2026, fiscal year ending March 31, 2027). A kind the title does not name is taken from a "Reporting period:" line at the top. After that, a period of the same kind that differs is reported where the text speaks of the current results: in a table heading cell, and in a sentence or heading with "results" or "this quarter". A cell, sentence or section heading with a word for a comparison or a later period ("prior year", "a year earlier", "compared with", "forecast", "outlook") is not compared, and neither is a period written next to the document''s own (| FY2025 | FY2026 |). A document whose title names its year with a marker ("Tax Year 2026", "the 2026 tax year"), such as an annual tax statement, is read even when it is not a results release, and a year with the same marker that differs anywhere in the body ("tax year 2025") is reported.'
+
+message:
+  ja: 「{label}」は、題の期間「{stated}」と違います
+  en: '"{label}" is not the period the title reports, "{stated}"'
+
+how_to_fix:
+  ja: 期間の名前を題に合わせて直してください。比べる相手の期間なら、「前年同期」「前期」のようにそう分かる書き方にしてください。
+  en: Correct the period name to the one in the title. If it is a period being compared with, say so ("a year earlier", "prior year").
+
+levels: { strict: error, normal: warning, relaxed: info }
+
+use_for: [business, technical, blog, legal, docs, academic, literature, speech]
+off_for:
+  literature: "Length, rhythm, repetition and register are the author's choices in literature, and dialogue mixes styles. Only what one work can hold itself to runs: consistent spacing and commas, and numbering without gaps."
+
+group: facts
+summary:
+  ja: 第2四半期決算の表の「当第3四半期」、令和8年分の明細の「2025年分」のような、題と同じ種類の別の期間
+  en: "Not the title's period: Q3 in a Q2 release, tax year 2025 in a 2026 statement"
+example:
+  ja:
+    before: |
+      # 2027年3月期 第2四半期決算短信
+
+      | 項目 | 前年同期 | 当第3四半期累計 |
+      | --- | --- | --- |
+      | 売上高 | 2,400百万円 | 2,640百万円 |
+    after: |
+      # 2027年3月期 第2四半期決算短信
+
+      | 項目 | 前年同期 | 当第2四半期累計 |
+      | --- | --- | --- |
+      | 売上高 | 2,400百万円 | 2,640百万円 |
+  en:
+    before: |
+      # Consolidated Financial Results for FY2026
+
+      Consolidated results for FY2025 are shown below.
+    after: |
+      # Consolidated Financial Results for FY2026
+
+      Consolidated results for FY2026 are shown below.
+
+not_flagged:
+  ja: 題に「決算」「業績」などの語と期間の名前が無い文書は読みません。前年同期、前期、前年同四半期、前回、比べ、予想、見通し、計画、次期のある升、文、節の中の期間、文書の期間と並べて書いた期間は比べません。第2四半期累計の文書の「第1四半期」のように文書の期間より前の四半期や半期は、累計の中の一部のことがあるので、「当第1四半期」のように今の期間の語が付くときだけ指します。2027年3月期と2026年度のように書き方の形が違う期間どうしは比べません。年分の文書では、「2025年」「2025年度」のように「年分」の付かない年、前年、比較、以前、繰越のある文、升、節の中の年、「分担金」のように「年分」が別の語の一部のときは比べません。
+  en: 'A document whose title has no results word and period name is not read. A period in a cell, sentence or section with a word for a comparison or a later period ("prior", "previous", "a year earlier", "compared with", "forecast", "outlook", "expect", "next year"), and one written next to the document''s own period, are not compared. A quarter or half before the document''s (Q1 in a six-month release) may be a part of the period and is reported only with a word for the current period ("this quarter"). A fiscal year named by its end date and one named by its start (2027年3月期 and 2026年度) are not compared with each other. In a document whose title names a tax year, a year without the same marker ("compared with 2025", "in 2025"), and one in a sentence, cell or section with a comparison word ("prior year", "last year", "or earlier", "carried forward"), are not compared. "For the year 2026" is not read as a tax year: annual reports use it for any year.'
+level_meaning:
+  ja: 題の決算期間と同じ種類の別の期間を、今の業績を言う所に書いたとき指摘
+  en: reported when a period of the title's kind but another value names the current results

package/rules/period-multiple-mismatch.yaml

@@ -0,0 +1,61 @@
+id: period-multiple-mismatch
+layer: L3
+status: stable
+how_to_find: period-multiple-mismatch
+# The amounts are read with currency-notation and amount-multiplier; the base labels are period-multiple-label, the words
+# period-multiple-word, counts written as words number-word (digit), and a range of counts range-connector.
+severity: warning
+
+name:
+  ja: 月払の何か月分と合わない額
+  en: An amount that is not the stated number of payments
+
+why:
+  ja: 保険や料金の案内で「年払（月払の12か月分）29,800円」のように、ある額を別の額の何か月分・何倍と書くと、読み手は計算せずにそのまま信じます。月払の額を書き直したのに年払の額が残ると、食い違います。「月払の12か月分」「月額の2倍」のように、もとの額の見出しと数が並んで書いてあり、その横に金額が一つだけあるとき、その金額を、前に書いた「月払 2,400円」のような額のその数倍と比べます。年払は一度に払うぶん月払の12倍より安いことが多いので、「月払の何か月分」と書いていない年払は比べません。「9.8万円」は書いた桁までの丸めを認めます。
+  en: 'When a premium table says "Annually (12 monthly payments) $298", readers trust the figure without working it out. It goes wrong when the monthly premium is changed and the annual line is not. When a count is stated as a number of a base ("12 monthly payments", "12 times the monthly premium") with one amount beside it, the amount is compared with that many of the base stated before it ("Monthly: $24"). An annual premium is often cheaper than twelve monthly ones, so an annual amount that does not say it is twelve monthly payments is never compared.'
+
+message:
+  ja: 「{multiple}」とある {amount} が、{label} {base} の{count}倍と合いません
+  en: "{amount} is not {count} × the {label} amount ({base}), though it says {multiple}"
+
+how_to_fix:
+  ja: もとの額・数・金額のどれが正しいかを確かめて、ほかを直してください。割引や税を含めた額なら、そう書いてください。
+  en: Check which of the base amount, the count and the amount is right, and correct the others. If the amount has a discount or tax in it, say so.
+
+levels: { strict: error, normal: warning, relaxed: info }
+
+use_for: [business, technical, blog, legal, docs, academic, literature, speech]
+off_for:
+  literature: "Length, rhythm, repetition and register are the author's choices in literature, and dialogue mixes styles. Only what one work can hold itself to runs: consistent spacing and commas, and numbering without gaps."
+
+group: facts
+summary:
+  ja: 「月払の12か月分」とあるのに、月払の12倍と合わない金額
+  en: An amount said to be 12 monthly payments that is not 12 times the monthly one
+example:
+  ja:
+    before: |
+      | 払込方法 | 保険料 |
+      | --- | --- |
+      | 月払 | 2,400円 |
+      | 年払（月払の12か月分） | 29,800円 |
+    after: |
+      | 払込方法 | 保険料 |
+      | --- | --- |
+      | 月払 | 2,400円 |
+      | 年払（月払の12か月分） | 28,800円 |
+  en:
+    before: |
+      | How you pay | Premium |
+      | --- | --- |
+      | Monthly | $24 |
+      | Annually (12 monthly payments) | $298 |
+    after: |
+      | How you pay | Premium |
+      | --- | --- |
+      | Monthly | $24 |
+      | Annually (12 monthly payments) | $288 |
+
+not_flagged:
+  ja: 「年払 28,000円」のように何か月分と書いていない額、「11〜12か月分」のような数の幅、一つの文や行に金額が二つ以上ある記載、割引・税込・約・各などの語がある記載、もとの額の見出しに金額が二つある記載は比べません。もとの額より前にある記載は、文書のもとの額が一つのときだけ比べます。通貨の違う額は比べません。表の列見出しだけに「月払」とある表は読みません。
+  en: 'An annual amount that does not say it is a number of payments ("Annually $280"), a range of counts ("11 to 12 payments"), a sentence or row with two amounts, a statement with a discount, tax, "about" or "each", and a base label with two amounts are not compared. A multiple before any base is compared only when the document states one base. Amounts in different currencies are not compared. A table that names the monthly amount only in its column heading is not read.'