Mashrabaمشربة

النَّحْو وَالصَّرْفSyntax and morphology

al-naḥw wa-l-ṣarf

Classical grammar has two halves. Naḥw studies how words behave in a sentence, above all their endings (iʿrāb). Ṣarf studies how a word is built: its root, its pattern, and what happens to weak letters. The reader's word cards show both: the grammar line (case, person, tense) is naḥw; the pattern (wazn) and Form are ṣarf.

Ibn Jinnī's definition

أما حدها فإنها أصوات يعبر بها كل قوم عن أغراضهم

As for its definition: it is sounds by which every people expresses its purposes. al-Khaṣāʾiṣ, line 442 (vol. 1, p. 33) ↗

Where does language come from?

Three answers run through the tradition: language was taught by God (tawqīf, 'instruction'), agreed among people (iṣṭilāḥ or muwāḍaʿa, 'convention'), or some of both. Each voice below says which it holds.

  • Ibn Fārisd. 395/1004RevealedArabic was taught by God, beginning with Adam, and completed by the time of the Prophet.

    أقول: إن لغة العرب توقيف. ودليل ذلك قوله جل ثناؤه: وعلم أدم الأسماء كلها

    I say: the language of the Arabs is by instruction (tawqīf). The proof is His word, glorious be His praise: 'And He taught Adam the names, all of them' (Q 2:31). al-Ṣāḥibī, lines 83–84 (vol. 1, p. 1) ↗

    The digital text reads أدم for آدم; quoted as it stands.

    بل وقف الله جل وعز آدم عليه السلام على ما شاء أن يعلمه إياه مما احتاج إلى علمه في زمانه

    Rather, God, mighty and glorious, instructed Adam, peace be on him, in what He willed to teach him, of what he needed to know in his time. al-Ṣāḥibī, lines 103–105 (vol. 1, p. 2) ↗

    Ibn Fāris (d. 395/1004), al-Ṣāḥibī fī fiqh al-lugha, print edition not recorded by OpenITI. Digital text: 0395IbnFarisQazwini.Sahibi.JK010092-ara1 (OpenITI RELEASE v2025.1.9). CC BY-NC-SA 4.0 (OpenITI digital text); the classical text is public domain, the modern edition is not.

  • Ibn Jinnīd. 392/1002UndecidedMost thinkers hold convention; his teacher said revelation; he finds the imitation of natural sounds a sound view, and ends suspended between the two.

    أكثر أهل النظر على أن أصل اللغة إنما هو تواضع واصطلاح لا وحي وتوقيف

    Most of those who reason hold that the origin of language is only agreement and convention, not revelation and instruction. al-Khaṣāʾiṣ, lines 544–545 (vol. 1, p. 40) ↗

    وذهب بعضهم إلى أن أصل اللغات كلها إنما هو من الأصوات المسوعات كدوى الريح وحنين الرعد وخرير الماء

    Some held that the origin of all languages is in the sounds that are heard: the howl of the wind, the rumble of thunder, the murmur of water. al-Khaṣāʾiṣ, lines 625–628 (vol. 1, p. 46) ↗

    The digital text reads المسوعات for المسموعات; quoted as it stands.

    وهذا عندي وجه صالح ومذهب متقبل

    To me this is a sound approach and an acceptable view. al-Khaṣāʾiṣ, lines 627–628 (vol. 1, p. 47) ↗

    فأقف بين تين الخلتين حسيرا وأكاثرهما فأنكفئ مكثورا

    So I stand between these two positions, worn out; I try to outweigh them, and turn back outweighed. al-Khaṣāʾiṣ, lines 640–642 (vol. 1, p. 47) ↗

    Ibn Jinnī (d. 392/1002), al-Khaṣāʾiṣ, ed. Muḥammad ʿAlī al-Najjār (Beirut: ʿĀlam al-Kutub). Digital text: 0392IbnJinniMawsili.Khasais.JK001302-ara1 (OpenITI RELEASE v2025.1.9). CC BY-NC-SA 4.0 (OpenITI digital text); the classical text is public domain, the modern edition is not.

  • al-Fārābīd. 339/950Convention, arising naturallyThe philosopher's account: people point, then make sounds; the sounds their bodies make most easily become their letters, and a word becomes a word when speaker and hearer settle on it.

    استعمل الإشارة أولا في الدلالة على ما كان يريد ممن يلتمس تفهيمه إذا كان من يلتمس تفهيمه بحيث يبصر إشارته، ثم استعمل بعد ذلك التصويت

    He first used pointing to show what he wanted to the one he sought to make understand, when that one could see him point; after that he used sounds. Kitāb al-Ḥurūf, lines 1562–1564 (vol. 1, p. 37) ↗

    ويكون ذلك هو السبب الأول في اختلاف ألسنة الأمم. فإن تلك التصويتات الأول هي الحروف المعجمة

    This is the first cause of the difference between the tongues of nations: those first sounds are the letters. Kitāb al-Ḥurūf, lines 1587–1588 (vol. 1, p. 37) ↗

    فيكونان قد اصطلحا وتواطئا على تلك اللفظة

    And so the two of them have agreed and settled on that word. Kitāb al-Ḥurūf, lines 1604–1605 (vol. 1, p. 38) ↗

    al-Fārābī (d. 339/950), Kitāb al-Ḥurūf, print edition not recorded by OpenITI; its paragraph numbers are kept in the text. Digital text: 0339AbuNasrFarabi.Huruf.JK010697-ara1 (OpenITI RELEASE v2025.1.9). CC BY-NC-SA 4.0 (OpenITI digital text); the classical text is public domain, the modern edition is not.

  • al-Suyūṭīd. 911/1505The map of positionsThe first chapter of the Muzhir gathers the debate. After quoting Ibn Fāris and Ibn Jinnī at length, he gives, from Fakhr al-Dīn al-Rāzī's al-Maḥṣūl, four positions, and says the careful scholars suspend judgement on all but one.

    الألفاظ إما أن تدل على المعاني بذواتها أو بوضع الله إياها أو بوضع الناس أو بكون البعض بوضع الله والباقي بوضع الناس

    Words signify meanings either by their very selves, or because God laid them down, or because people did, or some by God's laying-down and the rest by people's. al-Muzhir, lines 314–316 (vol. 1, p. 18) ↗

    والمحققون متوقفون في الكل إلا في مذهب عباد

    The careful scholars suspend judgement on all of these, except the view of ʿAbbād [which they reject]. al-Muzhir, line 321 (vol. 1, p. 18) ↗

    al-Suyūṭī (d. 911/1505), al-Muzhir fī ʿulūm al-lugha wa-anwāʿihā, ed. Fuʾād ʿAlī Manṣūr (Beirut: Dār al-Kutub al-ʿIlmiyya). Digital text: 0911Suyuti.MuzhirFiCulumLugha.Shamela0006936-ara1 (OpenITI RELEASE v2025.1.9). CC BY-NC-SA 4.0 (OpenITI digital text); the classical text is public domain, the modern edition is not.

Contested. Never settled. Ibn Fāris holds revelation; Ibn Jinnī reports a majority for convention and ends undecided; al-Fārābī describes convention arising among people; al-Suyūṭī, after al-Rāzī, records suspended judgement. The stakes were also theological: al-Suyūṭī notes that Ibn Fāris was a Sunnī and that Ibn Jinnī and his teacher Abū ʿAlī al-Fārisī were Muʿtazilīs.al-Muzhir, line 216 (vol. 1, p. 14) ↗al-Muzhir, line 218 (vol. 1, p. 14) ↗

Do sounds fit their meanings?A second question runs beside the first: is the link between a word's sound and its meaning arbitrary, or is there a fit? The grammarians found a fit in many words; one theologian made it a law of nature; most rejected that.
  • Ibn Jinnīd. 392/1002A fit in many wordsIn the chapter 'on words touching the likenesses of their meanings', the Arabs chose soft letters for soft things and hard for hard.

    فالخضم لأكل الرطب كالبطيخ والقثاء وما كان نحوهما من المأكول الرطب . والقضم للصلب اليابس

    Khaḍm is for eating what is moist, like melon and cucumber and moist food of that kind; qaḍm is for what is hard and dry. al-Khaṣāʾiṣ, lines 7904–7905 (vol. 2, p. 157) ↗

    فاختاروا الخاء لرخاوتها للرطب ، والقاف لصلابتها لليابس حذوا لمسموع الأصوات على محسوس الأحداث

    They chose the khāʾ, for its softness, for the moist, and the qāf, for its hardness, for the dry: modelling the heard sounds on the felt events. al-Khaṣāʾiṣ, lines 7908–7909 (vol. 2, p. 157) ↗

    النضح للماء ونحوه ، والنضخ أقوى من النضح

    Naḍḥ is for [sprinkling] water and the like; naḍkh [gushing] is stronger than naḍḥ. al-Khaṣāʾiṣ, lines 7910–7912 (vol. 2, p. 158) ↗

    وهو أن تتقارب الحروف لتقارب المعاني

    [In the chapter before:] that letters come close because meanings come close. al-Khaṣāʾiṣ, lines 7721–7722 (vol. 2, p. 146) ↗

    Ibn Jinnī (d. 392/1002), al-Khaṣāʾiṣ, ed. Muḥammad ʿAlī al-Najjār (Beirut: ʿĀlam al-Kutub). Digital text: 0392IbnJinniMawsili.Khasais.JK001302-ara1 (OpenITI RELEASE v2025.1.9). CC BY-NC-SA 4.0 (OpenITI digital text); the classical text is public domain, the modern edition is not.

  • ʿAbbād b. Sulaymān al-ṢaymarīMuʿtazilī, 3rd/9th c.A natural, necessary fitreportedKnown only through later reports, here al-Suyūṭī's: word and meaning are bound by a natural fitness that moved the word-maker. The majority rejected it; the language scholars, al-Suyūṭī adds, accepted a fit, but not a necessary one.

    بين اللفظ ومدلوله مناسبة طبيعية حاملة للواضع على أن يضع

    Between a word and what it signifies there is a natural fitness, which moved the word-maker to lay it down. al-Muzhir, lines 902–904 (vol. 1, p. 40) ↗

    لو ثبت ما قاله لاهتدى كل إنسان إلى كل لغة ولما صح وضع اللفظ للضدين

    If what he said were established, every person would find their way into every language, and a word could not be laid down for two opposites. al-Muzhir, lines 908–909 (vol. 1, p. 40) ↗

    وأما أهل اللغة والعربية فقد كادوا يطبقون على ثبوت المناسبة بين الألفاظ والمعاني لكن الفرق بين مذهبهم ومذهب عباد أن عبادا يراها ذاتية موجبة بخلافهم

    The scholars of language and Arabic, though, are almost agreed that there is a fitness between words and meanings; the difference from ʿAbbād is that he holds it essential and necessary, and they do not. al-Muzhir, lines 911–913 (vol. 1, p. 40) ↗

    al-Suyūṭī (d. 911/1505), al-Muzhir fī ʿulūm al-lugha wa-anwāʿihā, ed. Fuʾād ʿAlī Manṣūr (Beirut: Dār al-Kutub al-ʿIlmiyya). Digital text: 0911Suyuti.MuzhirFiCulumLugha.Shamela0006936-ara1 (OpenITI RELEASE v2025.1.9). CC BY-NC-SA 4.0 (OpenITI digital text); the classical text is public domain, the modern edition is not.

  • Plato, Cratylus4th c. BCNature or conventionparallel, not influenceThe same two answers, argued in Greek: Cratylus holds names correct by nature, Hermogenes by custom; Socrates hears motion in rho and gliding in lambda, and ends wanting likeness where it can be had, and convention where it cannot.

    ὀνόματος ὀρθότητα εἶναι ἑκάστῳ τῶν ὄντων φύσει πεφυκυῖαν

    [Cratylus says] that there is a correctness of name for each of the things that are, grown into it by nature. Cratylus, 383a ↗

    οὐ γὰρ φύσει ἑκάστῳ πεφυκέναι ὄνομα οὐδὲν οὐδενί, ἀλλὰ νόμῳ καὶ ἔθει τῶν ἐθισάντων τε καὶ καλούντων

    [Hermogenes:] No name belongs to anything by nature, but by the custom and habit of those who use it and call things by it. Cratylus, 384d ↗

    τὸ ῥῶ ἔμοιγε φαίνεται ὥσπερ ὄργανον εἶναι πάσης τῆς κινήσεως

    [Socrates:] The rho seems to me a kind of instrument for every motion. Cratylus, 426c ↗

    ὅτι δὲ ὀλισθάνει μάλιστα ἐν τῷ λάβδα ἡ γλῶττα κατιδών, ἀφομοιῶν ὠνόμασε τά τε λεῖα καὶ αὐτὸ τὸ ὀλισθάνειν

    Seeing that in the lambda the tongue glides most, [the name-giver] used it to make likenesses when he named the smooth, and gliding itself. Cratylus, 427b ↗

    ἀναγκαῖον δὲ ᾖ καὶ τῷ φορτικῷ τούτῳ προσχρῆσθαι, τῇ συνθήκῃ, εἰς ὀνομάτων ὀρθότητα

    [But] it may be necessary to make use of this vulgar thing too, convention, for the correctness of names. Cratylus, 435c ↗

    Plato (d. 348/347 BC), Cratylus, ed. John Burnet, Platonis Opera vol. 1 (Oxford: Clarendon Press, 1905; public domain). Digital text: urn:cts:greekLit:tlg0059.tlg005.perseus-grc2 (Perseus Digital Library, canonical-greekLit). CC BY-SA 4.0 (Perseus digitisation).

Contested. In both traditions. ʿAbbād's natural, necessary fit was rejected by most: then everyone would understand every language, and one word could not name opposites. What survived is the weaker claim, a fit in many words that the language-makers chose, which Ibn Jinnī illustrates and does not make a law.

Modern research. Modern work finds systematic but modest non-arbitrariness. It neither proves nor refutes the classical claims.

Findings summarised by us; each citation checked (title, venue, year) before listing.

English throughout: AI translation . The summaries, positions and “contested” lines are ours. Each voice is quoted in its own words unless marked reported (known only through a later author's report) or parallel, not influence.

In the classics4 books · why the rule holds · definitions
  • Qaṭr al-nadāIbn HishāmThe Word and Its Divisions
  • The AlfiyyaIbn MālikSpeech (al-kalām) and What It Is Composed Of (vv. 8–14)Morphology (at-Taṣrīf) (vv. 915–937)with Ibn ʿAqīl's commentary under each verse
  • al-ShāfiyaIbn al-ḤājibDefinition of morphology (taṣrīf)
  • Sībawayh, al-KitābSībawayhKnowledge of the parts of speech in Arabic
  • Why?al-Zajjājī, al-ĪḍāḥGrammar's reasons are not causes that compel; they are drawn from how the Arabs speak, and come in three kinds: for teaching, by analogy, and for debate. al-Khalīl compared himself to a man reasoning about a well-built house.
    تعليمية taʿlīmiyya
    for teaching: the rule as learned (“Zayd is nominative because he is the doer”)
    قياسية qiyāsiyya
    by analogy: why the rule holds (“inna governs because it resembles a transitive verb”)
    جدلية نظرية jadaliyya naẓariyya
    for debate and reflection: every further question about that analogy

    إن علل النحو ليست موجبة، وإنما هي مستنبطة أوضاعا ومقاييس

    The causes of grammar are not causes that compel; they are drawn out as conventions and measures. al-Īḍāḥ, line 506 (vol. 1, p. 64) ↗

    وعلل النحو بعد هذا على ثلاثة أضرب: علل تعليمية، وعلل قياسية، وعلل جدلية نظرية

    And the causes of grammar, after this, are of three kinds: causes for teaching, causes by analogy, and causes for debate and reflection. al-Īḍāḥ, lines 508–509 (vol. 1, p. 64) ↗

    فأما العلة القياسية فأن يقال لمن قال نصبت زيدا بإن، في قوله إن زيدا قائم: ولم وجب أن تنصب "إن" الاسم؟

    The analogical cause: one asks whoever said “I made Zayd accusative by inna” in inna Zaydan qāʾim: and why must inna make the noun accusative? al-Īḍāḥ, lines 519–520 (vol. 1, p. 64) ↗

    وأما العلة الجدلية النظرية فكل ما يعتل به في باب "إن" بعد هذا. مثل أن يقال: فمن أي جهة شابهت هذه الحروف الأفعال؟

    The cause for debate and reflection is every reason given in the chapter on inna beyond that, as when one asks: in what respect do these particles resemble verbs? al-Īḍāḥ, lines 527–528 (vol. 1, p. 65) ↗

    إن العرب نطقت على سجيتها وطباعها. وعرفت مواقع كلامها، وقام في عقولها علله، وإن لم ينقل ذلك عنها

    [al-Khalīl:] The Arabs spoke by their nature and disposition; they knew the places of their speech, and its causes stood in their minds, even if that was not handed down from them. al-Īḍāḥ, lines 550–551 (vol. 1, p. 66) ↗

    فمثلي في ذلك مثل رجل حكيم دخل دارا محكمة البناء؛ عجيبة النظم والأقسام

    My likeness in this is a wise man who enters a house of sound construction, wonderful in its order and its parts. al-Īḍāḥ, lines 552–553 (vol. 1, p. 66) ↗

    فجائز أن يكون الحكيم الباني للدار فعل ذلك للعلة التي ذكرها هذا الذي دخل الدار، وجائز أن يكون فعله لغير تلك العلة

    It may be that the wise builder of the house did it for the reason the visitor gave, and it may be that he did it for another. al-Īḍāḥ, lines 556–558 (vol. 1, p. 66) ↗

    English: AI translation The summary lines are ours.

    al-Zajjājī (d. 337/949), al-Īḍāḥ fī ʿilal al-naḥw, ed. Māzin al-Mubārak (Beirut: Dār al-Nafāʾis, 5th ed., 1406/1986). Digital text: 0337IbnIshaqZajjaji.IdahCilalNahw.Sham19Y0143170-ara1 (OpenITI RELEASE v2025.1.9). CC BY-NC-SA 4.0 (OpenITI digital text); the classical text is public domain, the modern edition is not.

  • al-Jurjānī, al-Taʿrīfāt · 3 definitions
    • النحو: هو علم بقوانين يعرف بها أحوال التراكيب العربية من الإعراب والبناء وغيرهما، وقيل: النحو: علم يعرف به أحوال الكلم من حيث الإعلال، وقيل: علم بأصول يعرف بها صحة الكلام وفساده.

      Grammar [al-naḥw]: a science of rules [qawānīn] by which the states of Arabic constructions are known with respect to declension [al-iʿrāb], indeclinability [al-bināʾ], and other matters. It is also said: grammar is a science by which the states of words are known with respect to weak-letter change [al-iʿlāl]. It is also said: a science of principles by which the correctness and incorrectness of speech are known.

      OpenITI · Taʿrīfāt, line 4858
    • الصرف: علم يعرف به أحوال الكلم من حيث الإعلال.

      Morphology [al-ṣarf]: a science by which the states of words are known with respect to weak-letter change [al-iʿlāl].

      OpenITI · Taʿrīfāt, line 2664
    • التصريف: هو علم بأصول يعرف بها أحوال أبنية الكلمة التي ليست بإعراب.

      Morphology [al-taṣrīf]: a science of principles by which are known the states of the forms of the word [abniyat al-kalima] that are other than declension [iʿrāb].

      OpenITI · Taʿrīfāt, line 1193

Arabic texts from OpenITI (CC BY-NC-SA 4.0); English titles and translations are AI. All the classics →

Further reading2 works