Learning right to left

Five of the languages Lingara works with are written right to left: Arabic, Sorani Kurdish, Persian, Dari and Pashto. Four of those are written in the Perso-Arabic script, and Sorani Kurdish uses an adapted form of it. This page is what that means in practice — both for reading the app in one of those languages and for learning one.

Two different things that get confused

There are two directions in play and they are independent.

The first is the language you read the app in. Choose Arabic, Sorani Kurdish, Persian, Dari or Pashto as your own language and the app flips: the document direction becomes right-to-left, text aligns to the right, navigation and rows mirror, and arrows in instructions point the other way. The website does the same thing — this page’s Arabic sibling is laid out mirror-image to the one you are reading.

The second is the language you are learning. That one does not flip the app. Learning Arabic from English leaves the interface left-to-right, because the interface is English: the buttons, the explanations, the settings and the error messages are all in the language you speak. What changes is that individual Arabic words and sentences inside that interface are rendered right-to-left, as a right-to-left run inside a left-to-right paragraph.

This is the correct behaviour and it is occasionally mistaken for a bug, so it is worth being explicit: a learner of Arabic who reads English should have an English app. Mirroring the whole layout because one field contains Arabic would move every control on the screen for the sake of one word.

What a mixed line actually does

Rendering an Arabic phrase inside an English sentence is handled by the bidirectional algorithm your operating system already implements, and it is mostly invisible. The places where it becomes visible are worth recognising so they do not read as errors:

Punctuation at the boundary. A full stop or comma immediately after a right-to-left run can appear on the left of it rather than the right. This is the algorithm resolving a neutral character between two directions, and it is what a native reader expects.

Numerals. Digits run left to right even inside right-to-left text, in every one of these languages. A date or a quantity embedded in an Arabic sentence therefore reads in the opposite direction from the words around it. That is normal, not a rendering failure.

Latin inside the run. A product name — Lingara, Plus, Google Play — stays in Latin letters and stays left-to-right. Our own locale files keep these names untranslated on purpose, so they look like an island in the middle of a right-to-left line. They are supposed to.

Learning the script

Four of the five share the Perso-Arabic script, and the differences between them are exactly the kind of thing worth knowing at the start rather than discovering at level 4.

Letters are joined, and most letters have different shapes depending on whether they stand alone or sit at the beginning, middle or end of a word. A flashcard showing a word in isolation shows the isolated or connected forms of that word — which is how you will meet it in text, but it is not a letter-by-letter drill. Learning the shapes is a separate exercise and is worth doing early.

Short vowels are usually not written. Arabic, Persian, Dari and Pashto all omit them in ordinary text, so a written word underdetermines its pronunciation in a way that no amount of staring resolves. This is the strongest argument for keeping pronunciation guides on while you start: see reading a new script for the control and when to turn it off.

The alphabets are not identical. Persian adds letters Arabic does not have, Dari is Persian’s Afghan variety and shares its alphabet, Pashto adds several more again, and Sorani Kurdish adapts the script further — including, unusually for the family, writing its vowels. Progress in one gives you a real head start in the others, and it does not give you the others.

Dari and Persian are close, and they are listed separately. They share a script and much of a vocabulary and differ in pronunciation, register and a good deal of everyday usage. Lingara treats them as distinct target languages because a learner aiming at one is not well served by the other’s word choices.

The site and the app both flip

If you read this website in one of the five, the layout you are looking at is the mirror of the English one, produced from the same source rather than maintained separately. Every marketing page, every legal page and this guide family’s header behave the same way.

A detail with a reason behind it: the guides and docs are English only. There is no Arabic guides tree, and there will not be one until the guides are genuinely translated rather than machine-rendered. A right-to-left page that is a mechanical translation of an English one is worse for its reader than an English page they can choose to read, and it is worse for a search engine that has to decide what it is. The thirty-one translated languages on this site are the interface and the marketing copy, which are short, checkable and worth the cost.

What does not change

The scheduler does not care which direction a language is written in. A card’s ease factor, interval and next review date are computed identically for Arabic and for Finnish, and the daily deck is assembled the same way. The level picker is the same 1 to 9. Pashto, Persian, Dari and Sorani Kurdish all use the CEFR framework, so their levels 7, 8 and 9 are synthetic in the same way French’s are — see what a level means.

The tutor answers in the language you speak, which for most learners of these languages means the answer arrives left-to-right and the example it quotes arrives right-to-left, in the same paragraph. That is the mixed-line behaviour described above, working as intended.

If something genuinely renders wrongly — a control mirrored that should not be, a line broken in the wrong place, a word rendered in disconnected letters — that is a bug and we would like to see it. Email support@getlingara.com with a screenshot; the screenshot is worth more than the description, because bidirectional layout faults are very hard to state in words.