Target language and gloss language

Everything in Parseh is about one language and written in another, and the two need not be the one you would guess: an Italian learning English wants the meanings in Italian, not in English. So every book, every video and every studio document names both.

The first decides everything about the text: which script is looked for, which font and direction it is set in, which passes a reading edition has, which folder it is filed under, which dictionary is looked in, what the transliteration line is called. The second decides only what the meanings (or the notes) are set in: their direction, their hyphenation, what the reader calls them.

1. Books and videos#

You choose both when you add the thing:

Add a book

(/books/add/) has Language (“The language the book teaches — it sets the folder it is filed under, the fonts it is set in, and the conventions it is annotated by.”) and Glosses in (“The language the meanings are written in. An Italian learning Persian wants them in Italian.”). A note under the second says what the pair you chose means; for a left-to-right book the right-to-left glosses, Persian and Arabic, are greyed out, because a right-to-left gloss of a left-to-right text cannot have its vocabulary lines set in order.

Add a video

(/youtube/add/) has Language and Glossed in, with English chosen; the page remembers your gloss language for the next video.

Both selects of the gloss list the eleven languages Parseh teaches, then the languages it only writes in: Portuguese, Dutch, Catalan, Romanian, Polish, Russian, British English and American English. A book glossed in Portuguese gets its meanings in Portuguese, set and hyphenated as Portuguese — and nothing else, because Portuguese has no folder, no fonts and no note types here: nobody studies it in Parseh.

Neither choice is changed later from the book info or video info sheet, which edit titles, authors and the like. A book or a video that does not say what it is glossed in is glossed in English, which is what everything made before the choice existed means — nothing had to be changed for it.

A book or video glossed in the language it teaches — English glossed in English, Italian in Italian — is a monolingual edition: its meanings are definitions rather than translations. That is the book’s choice, not a property of the language, and Parseh treats it as such (the add page says so).

What the gloss language changes elsewhere#

The labels.

In a book’s chunk sheet the box a meaning is typed into is labelled with the gloss language (italian), and a card’s two sides are named after both languages (Persian → English); in a monolingual edition they become meaning, and word → meaning.

The reading help.

A corpus and a translation model are pairs: the reader asks for the one matching its own gloss language, so a Persian book glossed in Italian needs a Persian–Italian corpus, and can have no model at all, since Mozilla’s pairs all go through English. The dictionaries explain every word in English whatever the gloss language, and a verb entry the dictionary drafts for a book not glossed in English leaves the meaning empty for you to write.

Anki cards

carry the gloss language with them, so an Italian or an Arabic meaning shows as itself, in its own direction.

2. Studio documents#

A studio document says the same two things at its top:

---title: The particle はtarget: jalang: it---

target:

is the language being learned, a registry code. Without one, a document is about Persian — Parseh’s first language, and what everything written before the others arrived was about. A new document starts in the language the language chips are set to, unless they are set to all. The document is filed in the library under that language, and moved there when a save changes it.

lang:

is the language the notes are written in. It is not cosmetic: it chooses the hyphenation of the PDF. Without one, the document is set as English. It takes the same codes a gloss does. It also sets which way the editor’s source box runs to begin with — right to left for lang: fa or lang: ar — until you press ⇤ RTL editor or ⇥ LTR editor for that document.

A Japanese note written in Italian has target: ja and lang: it. When you bring in a Markdown file whose header lacks them, the studio asks in a dialog — Written in and The language it is about, a select each, the second set to the chip’s language when one is picked — and adds the lines it was missing.

The guide you are reading works the same way: each page names its target, Persian by default, and each parseh-example may name its own:

---target: eslang: en---[El libro]{tl} = *the book* — a Spanish example on a page whose otherexamples are Persian.

3. Why the keys are called fa and en#

If you ever open a card file, an annotations.json or a chapter’s .tex, you will find the text under the key fa and the meaning under en — whatever the languages. They are the names of two slots, left from the days when Persian and English were the only two languages: fa is the text in the language being learned, and en is the meaning, in the gloss language. An en holding Italian is not a mistake; it is a book whose gloss is it. Renaming them would have meant rewriting every book, video and card, to say what "gloss" already says.