The verb has ten shapes and each of them means something predictable. Take a root and put it into form II and you get an intensified or a causative version; form III turns the action towards somebody; form V is the reflexive of form II; form VII makes it happen to the subject; form X means seeking or asking for it. So kataba is he wrote and istaktaba is he asked somebody to write, and nothing about that had to be memorised as a separate word.
That system is the whole of Arabic grammar worth front loading. Learn the ten forms as meanings rather than as tables and a large part of the language becomes derivable instead of collectible.
Six patterns that carry the grammar
- kataba is the citation form. Dictionaries list the third person masculine past, so the word you look up means he wrote rather than to write.
- there is no verb to be in the present. Al-bayt kabir is the house is big, two words and no verb between them, and the sentence is complete.
- kitab al-walad, the boy's book, literally book the-boy. Possession is built by putting two nouns together, and the first one loses its article to show it.
- al-kitab al-jadid, the new book. An adjective follows its noun and repeats the article, so definiteness is marked twice.
- the dual: kitaban, two books. Arabic has a separate number for exactly two, on nouns, adjectives and verbs alike.
- laysa negates a nominal sentence, while lam and ma negate verbs in different tenses. Negation is several words rather than one, chosen by what is being denied.
Reading is an inference and it needs both halves
An unvocalised word gives you consonants and nothing else, so reading it means running two identifications at once. Which three letters are the root, and which pattern the word has been poured into. Get both and the vowels arrive with the pattern; get one and the word stays ambiguous, because the same consonants in different patterns are different words.
The sentence supplies the third input, and it is the one learners forget to use. If the surrounding words require a place, the pattern is probably the one that makes places. If they require an agent, it is probably the one that makes agents. Fluent reading is not slower decoding, it is this three way guess completing before the eye moves on, and it is why an experienced reader can read a text and a beginner cannot read the same text more slowly.
There is one further thing Arabic does that English does not, and it is a relief. A present tense sentence has no verb to be at all: the equivalent of the house is big is simply the house big. Nothing is missing and nothing has to be learned.
The case endings are drilled and then never seen
Here is where a term disappears. Standard Arabic has a case system marked by short vowel endings, and courses teach it thoroughly because it is what the grammatical tradition is built on. Those endings are written only in vocalised text, which is not what you will read, and they are dropped in almost all speech, including formal speech, including by educated native speakers on television.
So a learner spends months on a system they will neither see nor hear, and often arrives at the point of trying to speak with the endings still attached, which sounds like somebody reciting. The honest allocation is to learn to recognise the endings, because they appear in vocalised text and in recitation and in poetry, and to stop well short of producing them automatically. What that time buys instead is the ten forms and the broken plurals, both of which turn up in every sentence you will ever read. Deciding where hours go when a syllabus and a language disagree is the general problem in turning a vague intention into hours on a calendar.
Three inputs, one reading
What holds up a reading of an unvocalised word
Remove any column and the sentence above stops being true, which is why speed here comes from all three arriving at once.
- The root, identifiedThree consonants pulled out of the letters on the page
- The pattern, recognisedWhich shape the root was poured into, and the vowels it carries
- The sentence around itWhich tells you whether a place, an agent or an act is wanted
Fluent reading is this guess completing before the eye moves, not a slower version of decoding.
The third column is the one learners leave out, and it is free. A word that is ambiguous on its own is very often unambiguous in a clause, and using the clause is the difference between reading and puzzling.
Two tools that name the root and the form for you
The Quranic Arabic Corpus gives a full grammatical analysis of every word in its text: the root, the form, the part of speech and the case, all free and all linked. It replaced working out the analysis by hand and having no way to check it, which is the position most learners are in for the first year. The text is religious and the tool is a linguistic one, and it is the largest free corpus of analysed classical Arabic there is.
A model does the same job on arbitrary text, and this is the one place it is genuinely reliable. Paste a word and ask for the root and the form number. It replaced flipping through a dictionary organised by root before you can identify the root, which is the circular problem every beginner meets. Free at this level. Ask for the analysis rather than the translation, and check the root against a dictionary: models are strong on morphology and weaker on rare vocabulary, and the failure is silent.
The mistake: learning the ten forms as conjugation tables
The forms are usually presented as ten sets of endings to be memorised, which is a great deal of table for very little return, because the endings are regular and the interesting information is in what each form does to the meaning.
Learn them as meanings with one example each, and the tables follow on their own from a few hundred encounters. Form II intensifies or causes. Form III does it towards or with somebody. Form V undoes form II onto the subject. Form VII is what happened to it. Form VIII is doing it to yourself. Form X is seeking it. Six of the ten cover almost everything you will meet, and a learner who owns those six can guess at a verb they have never seen and be right more often than not, which is the return the tables do not offer.
What the patterns cannot supply
The system generates a great deal and there is a hole in it, which is the plural of nouns: many of them are broken plurals that follow no derivable rule and have to be stored with the word, and that is learning by root instead of by word. Everything above describes the written standard, and the language people speak has a different negation, a different tense marker and different question words, which is which to learn first, and what each is used for. Reading any of it depends on the script being automatic, in right to left, joined letters, and vowels you infer, and hearing which form somebody used, at speed, is news Arabic, and the Arabic people actually speak.