What differs, language by language
The doors, the buttons and the file formats are the same for every language. This page is everything that is not: what each language needs to be read, and what each part of Parseh does about it.
1. The levels of a reading edition#
A reading edition gives each passage several times — as levels — and how many, and what each one is, depends on the language. The reader shows one button per level in its header, named as below and captioned levels; each button’s tooltip says what the level shows, and turning one off hides that level. The names are each language’s own defaults, and a person can change them (Levels and their names).
| Language | Its levels, by the name on their button |
|---|---|
| Persian | With vowels the vowelled attempt · Chunks chunks and glosses · Plain bare naskh · Nastaliq nastaliq |
| Arabic | With vowels the vowelled attempt · Chunks chunks and glosses · Plain unvowelled, as Arabic is written |
| Japanese | Furigana with furigana over each word · Kana only the reading alone, in kana · Chunks chunks and glosses · Plain plain, as Japanese is written · Vertical vertical (tategaki) |
| Chinese | Pinyin with pinyin over each word · Pinyin only the reading alone, in pinyin · Chunks chunks and glosses · Plain plain, as Chinese is written · Vertical vertical |
| Italian, French, German, Turkish, English, Spanish, Hindi | Sentence the sentence · Chunks chunks and glosses |
Why these and not others:
- A plain level exists only where it shows something the first does not.
- Persian and Arabic are written in the reading editions with their vowel marks (harakat) — the first level, With vowels, is the vowelled attempt — and Plain takes them off, as the languages are normally printed. Japanese and Chinese take away the readings over the words. The Latin-script languages and Hindi have nothing to take off (Devanagari writes its vowels), so their first level, Sentence, already shows all a plain level could.
- A fourth face
- is Persian’s alone: the passage again in nastaliq, the calligraphic hand of Persian poetry and titles.
- Japanese and Chinese have a reading level and a vertical level.
- In the first level each word of a chunk wears its reading over it, where the chunk carries its word line (a Japanese chunk without one wears its kana over the whole of it); Kana only (Pinyin only) is the reading alone — the passage in kana, or in pinyin, as it is said, with the punctuation that ends each chunk — and Vertical sets the passage in columns read top to bottom and right to left.
- Persian and Arabic can put the marks away.
- Diacritics, in the ⚙ panel, takes the very marks the plain level leaves off from the first two levels, and from a video’s lines — on the screen only (Levels and their names).
2. Readings, words and columns#
| Japanese | Chinese | the rest | |
|---|---|---|---|
| Transliteration line | rōmaji (Hepburn) | pinyin, with tone marks | see below |
| A reading beside it | kana: every glossed chunk carries it, and the checkers require it | none — pinyin is the transliteration | none |
| Words | every chunk may carry a word line, 山(やま) へ 柴刈り(しばかり) に: its words and each word’s reading | the same, with pinyin: 我(wǒ) 想(xiǎng) 喝(hē) 茶(chá) | no need: spaces already divide the words |
| Vertical text | the Vertical level; a studio block with vertical | the same | — |
| No word separator | a chunk counts as one word everywhere; the dictionary cuts it (or follows its word line) | the same | — |
The word line is what the furigana of the first level stand over, what the dictionary looks up word by word, and what the cloud’s 山: I know this buttons work by: a word you know stops wearing its reading wherever it appears in that book or video. The video player of a Japanese or Chinese video has one more button, named after the reading — kana only or pinyin only — which shows every phrase as it is said, in place of its text.
3. The transliteration, language by language#
Each language’s rules for writing its lines live in its conventions file, docs/lang/<code>.md — the rules every prompt that asks a model to annotate that language includes, and the page to read when you wonder why a gloss is written the way it is.
| Language | The transliteration line |
|---|---|
| Persian | required. A scheme of its own: long ā, š č ž x q, the emphatics flattened, transparent morphology hyphenated (mi-kone, ketāb-hā); what is said, colloquial forms included. |
| Arabic | required. A scholarly scheme: ā ī ū, ʾ and ʿ, the emphatics with a dot under them. |
| Hindi | required, because the script hides what it does not say: the inherent a printed in every consonant and often silent — कमल is kamal. Every noun is glossed with m. or f. |
| Japanese | required: Hepburn rōmaji written from the kana (Tōkyō, particles wa e o as said). |
| Chinese | required: pinyin with tone marks, never numbers; the neutral tone unmarked. |
| Italian, Spanish | optional, and given only where the spelling misleads — Italian’s open and closed e and o and its stress, Spanish’s silent h. Most chunks need none. |
| German | optional, where the spelling misleads (vowel length, the two ch, stress). Capitals are spelling — a noun always has one — and a separable verb is glossed as the whole verb, however far apart its halves are. |
| Turkish | optional and rarely needed: the spelling is phonemic. The vocabulary line carries the morphology instead — a word’s stem and each suffix in order. |
| French | optional in principle, given on almost every chunk in practice: the spelling hides the sound (silent endings, nasal vowels, liaisons). A respelling, not IPA. |
| English | written for every chunk with a word in it, in IPA, General American: English spelling is the least reliable guide to sound of any here. |
4. Verb entries#
A vocabulary line gives a verb as a \vb: three forms, each with its sound, then the meaning — the same shape in every language. What the three forms are, and the two labels printed before the second and third, are each language’s:
| Language | The three forms | Printed labels |
|---|---|---|
| Persian | infinitive · present stem · past stem | pres. / past |
| Arabic | perfect (3rd m. sg.; its form number in the sound) · imperfect (3rd m. sg.) · masdar | impf. / masdar |
| Italian | infinitive · 1st sg. present · past participle | pres. / p.p. |
| Japanese | dictionary form · -masu stem · -te form | stem / -te |
| French | infinitive · 1st sg. present · past participle | pres. / p.p. |
| German | infinitive · preterite (3rd sg.) · past participle | pret. / p.p. |
| Turkish | infinitive · present in -iyor (3rd sg.) · aorist (3rd sg.) | pres. / aor. |
| English | plain form · past · past participle | past / p.p. |
| Hindi | infinitive · stem · perfective (m. sg.) | stem / perf. |
| Spanish | infinitive · 1st sg. present · 3rd sg. preterite | pres. / pret. |
| Chinese | verb · A了B (a separable verb only) · A不B (a verb with a complement only) | split / can’t |
A pair left blank is not printed at all — a Chinese verb has its split form or its can’t form, never both. Whatever else a language’s verbs need goes in one parenthesis after the meaning: to drive (er fährt; aux. sein). The chunk sheet’s \vb{}{}{}{}{}{}{} button names the three forms of the book’s language in its tooltip, and the dictionary’s verb entries, where a language’s recipe recognises a verb, arrive in exactly this shape (Dictionaries). Every one of the eleven languages has such a recipe.
5. In each door#
| Door | What changes with the language |
|---|---|
| Book reader | the levels above and their names; the direction and side of the chunk column; the digits of the labels; ruby over the words of Japanese and Chinese; the ⚙ panel’s first slider named after the language (Persian text size), and a Height of vertical columns slider for Japanese and Chinese; Diacritics for Persian and Arabic |
| Video player | the direction and face of the transcript; the kana line of a Japanese phrase above its rōmaji; the kana only / pinyin only reading button for Japanese and Chinese; Diacritics for Persian and Arabic |
| Studio | which text is found by its script and which must be marked; the fonts offered by font= (nastaliq for Persian, gothic for Japanese); vertical for Japanese and Chinese; a lemma heading’s reading field and a kana insert button for Japanese; the target-text editor’s button, ✎ Persian, ✎ Japanese… |
| Reading help | a dictionary for every language, with extra rules for Persian, Arabic, French, Italian, Japanese and Chinese; a corpus of Japanese or Chinese only after its dictionary; Decompose Kanji / Decompose Hanzi for Japanese and Chinese only; translation models only to or from English |
| Anki | a pair of note types per language — Frank Italian and Frank Italian Opposites, and so on (Persian’s first is Frank YouTube Persian) — whose first field is named after the language (Headword for English), with a Reading field for Japanese; tags <language>-book and <language>-youtube (farsi-… for Persian) |
| Videos, adding one | which captions count as plain (in a script language, a caption with none of its letters); the words for seconds, minutes and chapters in the transcript |
The chapters on each door have the details: this page only says what the language decides.