Standard Arabic is a written language that is also spoken on certain occasions, and every dialect is a spoken language that is rarely written. That is the whole distinction, and it tells you what each one is for. The standard is what you read, what you hear from a newsreader and what you write; a dialect is what anybody says to anybody, including the newsreader, the moment the camera stops.
Run both from the start and give each its own job. Doing them one after the other, standard first, is what produces a person who can read a leader column and cannot buy bread.
The four groups, and six differences that actually separate them
- Egyptian, Levantine, Gulf and Maghrebi, the four broad families. They are not accents of one spoken language; the distance between the ends of that list is comparable to the distance between Romance languages.
- Egyptian jim is a hard g, so the name Gamal is spelled with the letter that is j everywhere else. It is the fastest single marker of Egyptian speech.
- Levantine qaf becomes a glottal stop in the cities, so qalb, heart, arrives as albeh.
- shu, eish and eh all mean what, in Levantine, Gulf and Egyptian respectively, and it is the question word that places a speaker fastest.
- biddi is I want in Levantine, where Egyptian says ayez and standard Arabic says urid. Three unrelated words for the commonest verb in a conversation.
- Maghrebi is the hardest for others, with heavy vowel loss and a large French and Berber layer, and Moroccan speakers routinely switch towards Egyptian to be understood in the east.
The distance is not the same in every direction
The dialects are not accents of the standard and they are not equally far from it. Egyptian and Levantine share a great deal with each other and are reasonably close to the standard in vocabulary; a Levantine speaker and an Egyptian speaker converse without effort. Gulf varieties sit nearby. Maghrebi Arabic, from Morocco, Algeria and Tunisia, is far enough away that speakers from the east frequently do not follow it, and Moroccans often accommodate by moving towards Egyptian or the standard.
What differs is not pronunciation alone. The negation is built differently, the present tense takes a prefix that the standard does not have, the question words are different words, and a large part of the everyday vocabulary is simply not the standard's vocabulary. A learner who has only the standard is not listening to a strong accent, they are listening to a related language.
Which one to take is decided by where you are going and by nothing else, with one thumb on the scale: Egyptian has a century of film and television behind it, so it has by far the most material and is understood almost everywhere, and Levantine is the common second choice for the same reason on a smaller scale.
The month everything comes out mixed
Here is the stall, and it arrives around the point where both systems are half built. You reach for a word and get the standard one in a conversation, or a dialect negation in something you are writing, and what comes out belongs to neither. It is uncomfortable, it feels like evidence that running two was a mistake, and it is the moment most learners drop one of them.
The mixing itself is not the problem, and it is worth knowing that educated native speakers mix constantly: there is a whole middle register where standard vocabulary sits inside dialect grammar, and it is what an interview or a lecture actually sounds like. The difference is that they control the mixture and you do not yet. The exit is to assign registers deliberately rather than to abandon one: dialect for anything spoken to a person, standard for anything read or written, and no exceptions for a month. The control arrives from that separation and not from more study.
Assigning the jobs
What each half of the language is for
The columns are the assignment to make for the first month, deliberately and with no exceptions, until the two stop bleeding into each other.
- Standard, always4
- Anything you read
- Anything you write
- News, and the subtitles under a drama
- A formal speech, or a sermon
- Dialect, always4
- Anything said to a person
- Television drama and film
- Songs, and social media
- The taxi, the shop, the family
- Where they mix3
- An interview or a panel
- A lecture, and a classroom
- A speaker moving register mid sentence
The third column is not a failure of either system and it is where fluent speakers spend most of their time. It is also the last thing a learner can do, because controlling a mixture requires owning both of the things being mixed.
Two ways to get the spoken language
Kallimni Arabi is the Egyptian course that most serious learners end up using, and its existence is the point: proper graded material for a dialect is rare, and most dialects have nothing comparable. It replaced learning the spoken language by ear from friends, which works and takes years. Paid, and there are equivalents for Levantine of varying quality.
A model will convert a standard sentence into a named dialect, which is genuinely new since 2023 and genuinely useful for the specific job of seeing what changes. It replaced needing a native speaker present for every small question. Free at this level. The limitation is larger here than anywhere else in this pillar: models are trained overwhelmingly on standard Arabic, they drift back towards it, and they are noticeably weaker on Maghrebi than on Egyptian. Ask for a specific dialect by name, check the output against a recording, and never take it as authority for a variety it has barely seen.
The mistake: treating the dialect as broken standard
A learner who has spent a year on the standard hears a dialect and files it as careless: endings dropped, vowels changed, words that are not in the dictionary. The judgement feels like standards and it is a decision not to learn the language people speak.
None of it is degraded. The dialects are the languages that developed naturally and the standard is the one that was maintained deliberately, so the historical relationship runs the other way from the one the word dialect suggests. Treating the spoken forms as forms, learned on purpose, is the whole difference between a learner who can hold a conversation in three years and one who cannot, and the same principle applies to the reduced forms of any language, argued out in material just past your level, and why that is the sweet spot.
Where each half leads
The standard is what makes reading possible at all, since almost nothing is published in dialect, and its machinery is three letter roots, and patterns that generate the vocabulary. The dialect is what makes listening possible, since almost nothing spoken is standard, and that is news Arabic, and the Arabic people actually speak. The vocabulary of each overlaps the other rather than deriving from it, which doubles the size of the list and is dealt with in learning by root instead of by word. And several consonants are pronounced differently in each, which is the throat sounds, and the emphatic consonants.