The eleven languages

Parseh began as a Persian toolbox and now teaches eleven languages through the same doors, with the same buttons and the same file formats. Nothing about how you use it changes from one language to the next; what changes is what each language needs in order to be read.

1. The table#

The first table is in the registry’s order, which every list of languages in Parseh follows — the language chips, the add pages’ selects, the rows of the reading-help page.

The code

is what everything is filed under: "language" in a book’s or a video’s details, target: at the top of a studio document, the data-lang every page carries.

The name

labels the language on screen — the first slider of the Aa panel is called Persian or Japanese, not text — and names its folder: books/persian/, youtube/videos/japanese/, markdown/library/italian/, exercises/italian/.

Its own name

is what a language chip shows.

The transliteration line

is the second line of a gloss — how the chunk is said, in Latin letters — and each language calls it by its own name. Where it is required, the checkers refuse a glossed chunk without one; in a Latin-script language it is a pronunciation aid, given where the spelling misleads and left out where it does not.

The digits

are the ones chapter and paragraph labels are written in: a Persian book’s third chapter’s first paragraph is ۳.۱. The reader’s buttons stay in Latin digits: they are controls, not text.

2. Scripts that can be found, and scripts that must be marked#

In Persian, Arabic, Hindi, Japanese and Chinese, the language’s text can be told from English by its letters alone. A studio document or a guide page finds it on its own and sets it in the language’s face and direction; a paragraph with no Latin letters in it becomes a whole block of the language:

The word کتاب means *book*, and کتاب‌ها is its plural.من کتاب را خواندم، و بعد به خانه رفتم.

---target: hi---The word किताब means *book*; नमस्ते is how you greet somebody.

Italian, French, German, Turkish, English and Spanish are written in the same letters as the prose around them, so nothing can be found by its script: every stretch of them is marked, with […]{tl} (or the language’s code, or any mark such as a colour or a transliteration):

---target: it---[Il libro]{tl} = *the book*, and [la casa]{teal translit:ˈka.za} = *the house*.

Japanese and Chinese are found by their script too — Chinese’s set of characters is Japanese’s without the kana, so a Japanese sentence is never taken for Chinese — and have two things of their own: a reading over the characters, and columns.

---target: ja---[漢字]{kana:かんじ translit:kanji} = *Chinese characters*[春はあけぼの。やうやう白くなりゆく山ぎは、すこしあかりて、紫だちたる雲のほそくたなびきたる。]{tl vertical height=10}

---target: zh---[中文]{translit:zhōngwén} = *Chinese*[从前，山下有一个小村子。村子里住着一位老人和他的猫。]{tl vertical height=8}

The studio’s own pages (the Studio and Dialect sections of this guide) have every mark in detail.

3. Direction#

Persian and Arabic run right to left, everything else left to right — and every element of a language’s text carries its direction, so a Persian word inside an English sentence, an Arabic gloss, a Persian sentence over its English translation all come out the right way round. In a book, the chunk column sits on the language’s own side: the chunk on the right and its gloss on the left for Persian and Arabic, the other way round for the rest. The chrome around the text — buttons, sheets, the editor’s sources column — stays left to right whatever the book.

Japanese and Chinese can also be set vertically, in columns read top to bottom and right to left: a reading edition’s last pass, and a studio block marked vertical.

4. Fonts#

Four faces travel with Parseh, in lib/fonts/: Vazirmatn, Noto Nastaliq Urdu, Noto Naskh Arabic and Noto Serif Devanagari, all under the SIL Open Font License. An Anki deck built for Persian, Arabic or Hindi carries its language’s first face inside the package, so the cards look the same on a phone.

The CJK faces do not travel: a CJK font is tens of megabytes. Pages use the device’s own — Noto Serif CJK JP and SC where they are installed (the fonts-noto-cjk package on Debian and Ubuntu has both), the system’s mincho, gothic and song faces on a Mac, a phone or Windows. The PDF of a Japanese or Chinese book needs the named face on the machine that builds it. The installer checks, for Japanese and for Chinese separately, as soon as there is a book or a video in that language, and says what to install when the face is missing.

The Latin-script languages ask for no font at all. Their text is set in the same roman as the glosses, so a chunk and its gloss differ only by size and colour.

Which face a language is set in is never written into a page: every page reads it from a small stylesheet generated from the registry when Parseh starts, so changing a language’s face is a change to one table.

5. The one table#

Everything on this page — and everything a language is to Parseh: its folders, its passes, its labels, its verb forms, its Anki note types, its LaTeX settings — lives in one table, lib/languages.json, the language registry, with a second file beside it, config/languages.json, for the languages added on this machine (read after it, and kept by every update). Every tool reads it, and nothing else in Parseh keeps a list of languages, a script’s letters, a font name or a table of digits. That is why a twelfth language is a row of that table and two small files (Adding a language), and why what differs, language by language can be listed in full.