1.Introductory observations

The most famous scholarly work on Arabic grammatical theory is Sībawayhi’s al-Kitāb (d. 161–94 AH/777-810 A.D.), which has been an object of examination by scholars from the time it was written, long before it became an authoritative source, to the present. Al-Kitāb is of a descriptive nature and is an examination of the Arabic language as it was used in the 2nd/8th century. Sībawayhi’s deductions were exploratory in nature, often ambiguous and vague. However, the prescriptive nature of an instructional manual does not require more than pedagogical explanations; therefore, teaching brief and concise material becomes more effective. Comparing instructional manuals such as al-Zaǧǧāǧī’s Ǧumal (d.337–340/948–951) and Ibn Bābašāḏ’s Muqaddima (d.469/1077) to Sībawayhi’s Kitāb, consequently, does not do justice to either type. How did these instructional manuals evolve? In what way do they differ, and where are they similar to al-Kitāb? What makes one more accessible to learners? These are some of the questions this study aims to answer through a careful examination of selected chapters from each type.

Scholarly investigation precedes writing instructional manuals, since investigative research leads to discovery which then evolves into educational materials. The science of linguistics is no exception. The linguist’s task is to analyse speech meticulously in order to describe the language examined and bring out the meaning behind its linguistic structure (Lehmann 1989Lehmann, Christian 1989 “Language Description and General Comparative Grammar”. Reference Grammars and Modern Linguistic Theory ed. by Gottfried Graustein & Gerhard Leitner, 133–162. Tübingen: Max Niemeyer. : 140). In many cases this descriptive analysis includes prescriptive elements, especially if the language under study is undergoing a process of standardization. Once these elements are thoroughly investigated, the basic theory of that particular language is formed (Dik 1989Dik, Simon C. 1989 “Functional Grammar and its Relevance to Grammar Writing”. Reference Grammars and Modern Linguistic Theory ed. by Gottfried Graustein & Gerhard Leitner, 33–55. Tübingen: Max Niemeyer. : 33). For many languages the investigative journey ends at this point. However, for the majority of languages that are investigated, the teaching of these languages is the next logical step, and different materials need to be developed, i.e., pedagogical or instructional manuals. Nevertheless, it is well known that descriptive grammars sometimes have been used for pedagogy. In this case teachers and learners must be made aware of the anomalies these descriptive grammars contain due to speculative approaches and theorizing of linguistic structures (Greenberg 1968Greenberg, Joseph H. 1968 Anthropological Linguistics: An introduction. New York: Random House.: 28). A specifically pedagogical grammar can only be possible after the descriptive research and the need for instruction are established. Descriptive analysis then becomes the basis for “pedagogical prescription” (Widdowson 1991Widdowson, Henry G. 1991 “The Description and Prescription of Language”. Georgetown University Round Table of Languages and Linguistics 1991 ed. by James E. Alatis, 11–24. Washington, D.C.: Georgetown University Press.: 12). The prescribed language in pedagogical grammars is based on the form of language chosen to be described. If standardization of a language is the aim of research, or more accurately, the result of such investigative research, then such a description ultimately becomes prescriptive, although not necessarily with this intention in mind when investigation starts.

Linguistic description and linguistic prescription are ultimately linked, and “the role of the linguist is still in large part prescription of a particular description” (Newmeyer 1978Newmeyer, Frederick J. 1978 “Prescriptive Grammar: A reappraisal”. Approaches to Language: Anthropological Issues ed. by William C. McCormack & Stephen A. Wurm, 581–593. The Hague: Mouton. : 585). Effectively, pedagogical grammars must first rely on accurate descriptive grammars. Observations attained from interpretation and examination of analysed data in a particular language must be reflected in pedagogy with simple direct rules that are ready to be understood by the learner of the language (Dirven 1989Dirven, René 1989 “Cognitive Linguistics and Pedagogic Grammar”. Reference Grammars and Modern Linguistic Theory ed. by Gottfried Graustein & Gerhard Leitner, 56–75. Tübingen: Max Niemeyer. : 58). The speculative nature of grammatical analysis does not belong in pedagogy, and pedagogy in languages cannot exist without linguistic description as “pedagogical grammars are ultimately translations of linguistic descriptions” (Tomlin 1994Tomlin, Russell S. 1994 “Functional Grammars, Pedagogical Grammars, and Communicative Language Teaching”. Perspectives on Pedagogical Grammar ed. by Terence Odlin, 140–178. Cambridge: Cambridge University Press. : 143). The need for such instructional materials is ultimately the driving force behind creating them, but, without descriptive grammars, they cannot exist.

In the case of Arabic, biographers report that the beginning of grammar was to preserve the language of the Quran from linguistic impurities that had infiltrated it through language contact as Islam and the Islamic empire expanded in the 1st/7th century (Baalbaki 2008Baalbaki, Ramzi 2008 The Legacy of the “Kitāb”: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: Brill. : 2–4). Arabs settled primarily in military camps within the newly conquered territories. As they mixed with the indigenous peoples of the conquered lands and non-Arabs began to embrace Islam, the Arabic language served as the language of the new empire: it became the means of official and unofficial communication. It was the language of government, culture, and the sciences. Arabic thus no longer belonged only to the people of Arabia. It became the lingua franca of the Islamic empire. Since Arabs born in the new territories, as well as indigenous people, used Arabic for communication, the language they used started to shift away from the Arabic of the peninsula due to language contact and language evolution. Scholars realized the challenge this posed to the understanding and correct recitation of the language of the Quran. Thus, the examination of the Arabic language began within the Quranic sciences in order to instruct people on the proper ways of reading Arabic. Accordingly, the study of grammar began to serve a pedagogical purpose.

The earliest stages of the development of Arabic grammatical theory remain obscure. Scholars have attempted to reconstruct this period, but with the scarce extant material, they cannot reach definitive conclusions. Consequently, the discussion begins with the famous scholarly work on Arabic grammar, al-Kitāb, composed by the Persian-born Sībawayhi (d.161–194/777–810),11.Abū Bišr ʿAmr b. ʿUṯmān b. Qanbar, known as Sībawayhi (d. 161-94/777-810), was Persian born and educated in Basra. For further details see Qifṭī (ʾInbāh II, pp. 346–360) and Sezgin (1984Sezgin, Fuat 1984 Geschichte des arabischen Schrifttums. Band IX: Grammatik. Leiden: E. J. Brill.: 51). which provides a descriptive, highly speculative analysis of the rules of Arabic grammar and essentially severs the tie it had with Quranic exegeses. By starting the investigation of Arabic grammar with Sībawayhi’s Kitāb, early pedagogical purposes for grammatical analysis are no longer noticeable in the descriptive language of the Kitāb itself. The initial interest for a pedagogical tool is lost in Sībawayhi’s theoretical approach (Baalbaki 2005Baalbaki, Ramzi 2005 “Theoretical Coherency versus Pedagogical Attainability: The conscious bias of Arab grammarians”. Alttagsleben und materielle Kultur in der arabischen Sprache und Literatur: Festschrift für Heinz Grotzfeld zum 70. Geburtstag ed. by Thomas Bauer & Ulrike Stehli-Werbeck, 39–68. Wiesbaden: Harrassowitz.: 40). The shift from a need to instruct learners of Arabic, in order to correctly recite the Quran, to the development of a highly speculative grammar, unfortunately, cannot be traced with the extant sources available.

As time passed, the dwindling segment of society who lived outside urban settlements and spoke “good” Arabic naturally passed away and thus this spoken form of the language died out. Arabic grammar could not remain a descriptive grammar. It transformed, by necessity, into a highly prescriptive grammar of a language that should be spoken and written in a certain way. All this translates into the growing need for instructional manuals in the way people ought to speak and write.

It has long been thought that Sībawayhi’s al-Kitāb is difficult to read and understand, especially because early Arab grammarians attested to this belief. According to al-ʾAḫfaš al-Ṣaġīr (d.315–316/928–929), al-Kitāb contained muštabah “obscure points” for scholars to istanbaṭ-a wa-naẓar-a “deduce and examine” (Baġdādī Ḫizāna, p. 372). Other scholars, such as al-Zaǧǧāǧ (d.310–16/922–928) and al-Sīrāfī (d.368/978), confirmed the complexity and difficulty in understanding al-Kitāb (Hārūn 1988Hārūn, ʿAbd al-Salām Muḥammad 1988 See Sībawayhi, Al-Kitāb.: 33). Sībawayhi’s al-Kitāb is an immense and even intimidating piece of work. He painstakingly described and analysed the language of the Arabs, interpreting data to make sense of the language, and in the process, descriptions of grammatical structures for Arabic emerged. Reaching these deductions, he needed to examine a multitude of examples, often contradictory, and provide a plausible explanation and rule for each point. Without appreciating this aspect of his contribution, one may lose sight of the nature of his work.

Sībawayhi was not only engaged in the practice of language description, his aim was the explanation of the language’s structure. He was not writing a pedagogical manual. According to Carter, Sībawayhi’s Kitāb “is no pedagogical treatise: indeed, it is so inclusively descriptive, accommodating all the variations and irregularities found in natural language (but not in the language of pedagogues), that it is completely unusable for teaching” (Carter 2005Carter, Michael G. 2005 “Sībawayhi.” Dictionary of Literary Biography Volume 311: Arabic Literary Culture, 500–925 ed. by Michael Cooperson and Shawkat M. Toorawa, 325–331. Detroit: Thomson Gale.: 329). Although scholars of Arabic grammar of the time studied al-Kitāb, it was by no means a manual for beginners. These scholars were interested in the speculative nature of early Arabic grammatical theory, and this fact is discernable throughout the history of Arabic grammar. However, in due course, Arab grammarians had no choice but to be prescriptivists, as they could no longer engage in descriptive grammar, because native speakers of Classical Arabic eventually died out and disappeared (Owens 1993Owens, Jonathan 1993 “Arabic Syntactic Theory”. Syntax: An international handbook of contemporary research ed. by Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld & Theo Vennemann, Vol. I, 208–215. Berlin: Walter de Gruyter.: 210). This does not mean that later grammarians’ focus on prescription caused them to consider instructional manuals. They were interested in the process of prescribing the way Arabic should be, yet remained speculative and theoretical. Baalbaki (2005)Baalbaki, Ramzi 2005 “Theoretical Coherency versus Pedagogical Attainability: The conscious bias of Arab grammarians”. Alttagsleben und materielle Kultur in der arabischen Sprache und Literatur: Festschrift für Heinz Grotzfeld zum 70. Geburtstag ed. by Thomas Bauer & Ulrike Stehli-Werbeck, 39–68. Wiesbaden: Harrassowitz. discusses how early Arab grammarians were “so heavily engaged in” speculative grammar that they “were hardly concerned with the pedagogical attainability of their interpretations and justifications.” Their application and defense of the theory was “at the expense of simple or straightforward explanations which would have made better pedagogical sense” (39–40). Grammarians were more inclined to justifying usage than simplifying for learners by avoiding speculative methods. The concern “of the grammarians for theoretical coherency” was “at the expense of pedagogical attainability” (43, 45).

The process of moving from Sībawayhi’s grammar to a more pedagogical approach took time. His observations needed to be adjusted by later grammarians in order to be presented in a form more easily understood by learners of Arabic. This process started early but took centuries to reach its peak (Carter 2005Carter, Michael G. 2005 “Sībawayhi.” Dictionary of Literary Biography Volume 311: Arabic Literary Culture, 500–925 ed. by Michael Cooperson and Shawkat M. Toorawa, 325–331. Detroit: Thomson Gale.: 329). Later grammarians, who were concerned with language instruction, wrote manuals primarily for beginners in order to fill the gap left by theoretical grammars. These grammar manuals that were written for instructional purposes did not and could not replace those books that were dedicated to theoretical grammars. Both sets of grammars existed throughout the Arabic grammatical tradition (Baalbaki 2005Baalbaki, Ramzi 2005 “Theoretical Coherency versus Pedagogical Attainability: The conscious bias of Arab grammarians”. Alttagsleben und materielle Kultur in der arabischen Sprache und Literatur: Festschrift für Heinz Grotzfeld zum 70. Geburtstag ed. by Thomas Bauer & Ulrike Stehli-Werbeck, 39–68. Wiesbaden: Harrassowitz.: 42); however, especially at earlier times, pedagogy was not able to escape theory. It took time for traces of theoretical grammar to slowly fade away from pedagogy, but this was necessary for the education and instruction of those who sought instruction in Arabic.

The process of shifting to a more pedagogical approach did not happen in an atmosphere clear of conflict. The translation movement and the development of the rational sciences were well under way in the 3rd/9th century (Versteegh 1977Versteegh, Cornelis H. M. 1977 Greek Elements in Arabic Linguistic Thinking. Leiden: E. J. Brill. : 117–118). Rational thinking and logic stimulated minds, resulting in fierce debates in all sciences, including grammar. Additionally, sources tell us that from the earliest stages of grammatical analysis, there were two schools of grammar: the Basran and the Kufan. Versteegh stresses the importance of acknowledging that there was underlying agreement between both schools, despite the exaggerated differences mentioned in the literature (ibid., 111–112). The rivalry between both schools reached its peak in Baghdad during the second half of the 3rd/9th century, where it became the center of grammatical thinking, and scholars were able to seek both traditions often studying with scholars from both camps.

Approximately fifteen to twenty years elapsed following the shift of the grammatical schools of Basra and Kufa to Baghdad and the rivalry between both schools had subsided, allowing scholars to study under scholars trained in both camps. This tolerant atmosphere along with the infusion of logic and philosophy in scholars’ reasoning reinforced the deductions and rationale behind grammatical rules. At this stage in the history of Arabic grammar, al-Zaǧǧāǧī22.ʾAbū al-Qāsim ʿAbd al-Raḥmān b. ʾIsḥāq al-Zaǧǧāǧī (d. 337-40/948-951) was also Persian born. He was educated in Baghdad and taught in Damascus towards the end of his life (Qifṭī ʾInbāh II, pp. 160–161, Sezgin 1984Sezgin, Fuat 1984 Geschichte des arabischen Schrifttums. Band IX: Grammatik. Leiden: E. J. Brill.: 88). emerges as one of the leading grammarians of the period, especially with his ʾĪḍāḥ 33. Al-ʿĪḍāḥ fī ʿilal al-naḥw is a philosophical debate on the rational motivation of linguistic rules in Arabic grammar. It is not a textbook dealing with grammatical rules, but it deals with the reasons behind the rules. The book is divided into 23 chapters followed by several pages dealing with additional issues. and Ǧumal.

Zaǧǧāǧī’s Ǧumal is a concise general introduction to Arabic grammar. Zaǧǧāǧī explains and simplifies grammatical terminology used by the grammarians of his era, asserting the need to taqrīb ʿalā al-mubtadiʾ “clarify for beginners” (Zaǧǧāǧī Ğumal, p. 90). Unfortunately, there is no introduction for the book; consequently, the reason behind writing it is based on statements he made in his book. However, it is clear that by the time of Ibn Ḫarūf in 6th/13th century, al-Ǧumal was used as a textbook for learners of Arabic. He believes that Zaǧǧāǧī had written his work for “beginners”, asserting that Zaǧǧāǧī waḍaʿ-a-hu li-l-mubtadiʾ-īna wa-ittakal-a fī bayān-i-hi ʿalā al-muʿallim-īna “wrote it down for beginners and relied on instructors to clearly explain it” (Šarḥ, p. 243). Later, in the 7th/14th century, Ibn ʾAbī al-Rabīʿ mentions that al-Ǧumal was taught from childhood, affirming that ʾaḫaḏ-a al-našʾat-u al-ṣiġār-u bi-ḥifẓ-i [al-Ǧumal] wa-ta-fahhum-i-hi “the young [generation] undertook the memorization and understanding of al-Ǧumal” (Basīṭ, p. 157).

Zaǧǧāǧī himself mentions throughout the book that al-Ǧumal is muḫtaṣar “condensed”; however, other scholars of the period held this against al-Ǧumal, criticizing Zaǧǧāǧī for his “exaggerated conciseness and condensation” (Baṭalyawsī Ḥulal, p. 57). The shift to instructional simplicity was not readily appreciated by all. Yet others, while acknowledging that Zaǧǧāǧī was brief and concise, note that he ʾittakal-a … f ī bayān-i hāḏā al-faṣl-i ʿalā al-muʿallim-i liʾann-a-hu iḫtaṣar-a ʿalā ʿādat-i-hi “relied on instructors to explain and clarify this section, since, as usual, he condensed [the material]” (Ibn Ḫarūf Šarḥ, p. 408).

By the 5th/10th century, the grammatical shift towards instructional manuals reached a pivotal stage in history with Ibn Bābašāḏ’s Muqaddima.44.ʾAbū Ḥasan Ṭāhir b. ʾAḥmad b. Bābašāḏ b. Dāwūd b. Sulaymān b. ʾIbrāhīm (d. 469/1077) was born in either Iraq or Egypt. Both editors of his commentary are in disagreement as to his birthplace, but they agree that he had received his education in both Baghdad and Egypt (Šarīf 1978Šarīf, Muḥammad ʾAbū al-Futūḥ 1978 See Ibn Bābašāḏ, Naḥwiyya.: 27; ʿAbd al-Karīm 1976–1977ʿAbd al-Karīm, Ḫālid 1976–1977 See Ibn Bābašāḏ, Muḥsiba.: 11). He settled in Egypt where he became a well-known government official and taught at Ibn al-ʿĀṣ mosque. He divides it into ten sections that thoroughly deal with the parts of speech, all the forms of inflection, operators, modifiers and finally orthography. Ibn Bābašāḏ explains in his commentary that he wrote his Muqaddima for the purpose of tashīl wa-tawṭiʾa “simplification and introduction” to Arabic grammar for beginners (Šarīf 1978Šarīf, Muḥammad ʾAbū al-Futūḥ 1978 See Ibn Bābašāḏ, Naḥwiyya.: 2:16). He describes how some of his contemporaries had renamed this book as al-Muqaddima al-Muḥsiba. ʿAbd al-Karīm explains that the meaning of al-Muḥsiba is al-Kāfiya, i.e., that which should suffice without the need to consult other grammatical books (1976–1977ʿAbd al-Karīm, Ḫālid 1976–1977 See Ibn Bābašāḏ, Muḥsiba.: 27). Nevertheless, his Muqaddima has been considered so short and concise that Ibn Bābašāḏ’s students asked him to give them a commentary on his work more than 30 years later. This commentary is widely known as Šarḥ al-Muqaddima al-Muḥsiba. Other grammarians also thought his Muqaddima too concise, motivating them to write their own commentaries.

According to Carter (1985)Carter, Michael G. 1985 “When Did the Arabic Word Naḥw First Come to Denote Grammar?” Language and Communication 5:4.265–272. in his examination of the term naḥw, the Muqaddima’s significance stems from the fact that its composition provided the definitive form of naḥw, because of the existence of three main elements. The first element is the highly developed indigenous tradition of grammatical analysis, which was documented in Sībawayhi’s Kitāb as the manner in which the people spoke and was studied and examined for centuries. The second element is what Carter calls “imported scientific methodology,” which came about with the translation movement and introduced the importance of presenting clear and precise definitions. The third is the new Islamic educational system represented by the establishment of the madrasa or ‘academy’. Institutionalized learning emerged at the beginning of the 11th century, and thus a systematic methodology for instruction was needed. After examining al-Muqaddima, I came to believe that, in addition to the historical elements mentioned by Carter, its simplified and concise presentation with its division into ten sections, makes it an exceptional instructional manual in its time.

In order to appreciate fully the gradual development of instructional manuals this paper examines the chapters on istiṯnāʾ “exception” using the particle ʾillā “except”, written by the aforementioned scholars in their respective books. Before I go into the details these grammarians offer, I present a simple illustration of grammatical rules of exception in Arabic. Exception involves three distinct elements: the exceptive particle, in this paper, we will only deal with the particle ʾillā “except”; the mustaṯnā “excepted noun” or the noun after ʾillā; and the mustaṯnā min-hu “noun excepted from” or the noun before ʾillā. The following example maps these three elements:

Here we can see the three exceptive elements: the exceptive particle ʾillā; the mustaṯnā “excepted noun” or the noun after ʾilla, which is the proper noun Zayd; and the mustaṯnā min-hu “noun excepted from” or the noun before ʾilla, which is ʾaḥad “anyone”, where Zayd is clearly not part of and is excluded from “anyone”.55.See Ḥasan 1968Ḥasan, ʿAbbās 1968 Al-Naḥw al-Wāfī. Vol. II. Cairo: Dār al-Maʿārif.: 292–337 for a detailed account for the rules of istiṯnāʾ.

2.The role of Sībawayhi’s Kitāb

Baalbaki (2008)Baalbaki, Ramzi 2008 The Legacy of the “Kitāb”: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: Brill. argues that Sībawayhi presents basic grammatical rules throughout his book to account for the majority of grammatical usage, and, at the same time, finding appropriate explanations for anomalies (134–135). Sībawayhi supports his examples to prove these rules and thus establishes what Baalbaki calls “Basic Rules”, which Sībawayhi applies throughout his book while at the same time he is able to explain any variation from these rules (ibid.; 2005: 43). This flexibility allows Sībawayhi to document anomalies without jeopardizing the rules deduced from the data. Baalbaki argues that Sībawayhi is “keen to uphold what we called ‘basic rules’ by maximizing the applicability of the norm and minimizing deviations which undermine it” (2008: 155). Baalbaki further elaborates on the early grammarians’ obsession with maintaining the concept of ‘basic rules’, with the use of taqdīr “suppletive insertion”, which “denotes the restoration of missing elements by the grammarians to explain various aspects of the construction” (Baalbaki 2005Baalbaki, Ramzi 2005 “Theoretical Coherency versus Pedagogical Attainability: The conscious bias of Arab grammarians”. Alttagsleben und materielle Kultur in der arabischen Sprache und Literatur: Festschrift für Heinz Grotzfeld zum 70. Geburtstag ed. by Thomas Bauer & Ulrike Stehli-Werbeck, 39–68. Wiesbaden: Harrassowitz.: 45) and modification of grammatical function, which allows for a shift in grammatical function without jeopardizing the “basic rule” (p. 55).

Sībawayhi’s presentation of the istiṯnāʾ follows this ‘basic rule principle’ by introducing general guidelines that govern the majority of cases, followed by explanations for those cases that do not follow these general rules. Sībawayhi covers istiṯnāʾ in several chapters, beginning with the basic rules. He states that the mustaṯnā, the noun after ʾillā, takes one of two options: (1) the case ending that the noun would normally receive without ʾillā, i.e., before introducing the particle, or (2) the accusative case in the tanwīn-naṣb construction.

Sībawayhi utilizes the phrase ʿišr-ūna dirham-an to explain the tanwīn-naṣb construction. He explains kamā taʿmal-u ʿišr-ūna fī-mā baʿd-a-hā ʾiḏā qulta ʿišr-ūna dirham-an “as the twenty operates on the dirham when you say twenty dirhams” (Kitāb II, p. 310).66.For a detailed discussion on the effect of ʿišr-ūna on dirham see Carter (1972)Carter, Michael G. 1972 “ ‘Twenty Dirhams’ in the Kitāb of Sībawayhi”. Bulletin of the School of Oriental and African Studies 35:3.485–496. . This convoluted explanation is Sībawahi’s way describing the tanwīn-naṣb construction.77.Later grammarians used the term tamyīz for specific elements of this structure. Sībawayhi’s grammatical theory involves elements acting upon each other in different ways. One type of operation involves the tanwīn-nasb “accusative” structure, where it represents non-agreement between its elements, occurs outside already complete sentences, and cannot form true annexation units. He uses the phrase ʿišr-una dirham-an “twenty dirhams” to represent this operation in his Kitāb (Carter 2004Carter, Michael G. 2004 Sībawayhi. London & New York: I. B. Tauris.: 91–92). Carter explains that the “tanwīn-naṣb structure was as much an identifiable type of syntactical unit as the more familiar ‘true’ annexation and subject-predicate constructions” (Carter 1972Carter, Michael G. 1972 “ ‘Twenty Dirhams’ in the Kitāb of Sībawayhi”. Bulletin of the School of Oriental and African Studies 35:3.485–496. : 495). In Sībawayhi’s theory this includes all verbal complements (cf. Carter 2004Carter, Michael G. 2004 Sībawayhi. London & New York: I. B. Tauris.: 91). Unfortunately, later grammarians did not preserve this category, and, consequently, it has been lost to the theory.

To illustrate the first option, Sībawayhi explains that ʾillā and the mustaṯnā “noun that follows” are introduced into an utterance to establish the existence or affirm the situation that has been excluded or negated in the utterance before ʾillā. His statement tudḫil-u al-ism-a fī šayʾ-in tanfī ʿan-hu mā siwā-hu “You introduce the [excepted] noun to where everything else has been excluded” (Kitāb II, p. 310), also indicates that Sībawayhi means that the sentences he is considering are negated, proven by the examples he gives (ibid.):

The sentence structure before ʾillā in the above examples reflects negated sentences, and the function of the particle ʾillā is to exclude what follows from these negated structures. Thus, Zayd did come, did meet with, and did pass by the speaker. The nouns after the particle ʾillā all receive the appropriate grammatical inflections they would have received if the particle was never introduced into each sentence: Zayd in the first sentence is in the nominative case as the subject; Zayd is in the accusative case as the direct object in the second sentence; and Zayd is in the genitive case in the third sentence as the object of a preposition.

Sībawayhi clarifies tuǧrī al-ism-a maǧrāh-u … wa-lākinn-a-ka ʾadḫalta ʾillā li-tūǧib-a al-ʾafʿāl-a li-hāḏihi al-asmāʾ-i wa-li-tanfiy-a mā siwā-hā “You give the noun its rightful place [in the sentence structure] … however you introduce ʾillā to affirm the verbs for these [excepted] nouns and exclude others” (Kitāb II, p. 310). The only reason ʾillā is used is to deactivate the negation in the sentence structure just before it is used, and thus, affirming that Zayd did come, did meet with, and did pass by the speaker. The reason that this is allowed is that the main verbs in these clauses do not have all their arguments: the verb ʾatā-ni “came to me” is missing the subject Zayd. The verb laqītu “I met” is missing the object Zayd, and the verb marartu bi “I passed by” is missing the object of the preposition Zayd. These missing arguments are needed for these sentences to be complete, and they are not complete sentences, because they do not have the mustaṯnā min-hu “the noun before ʾillā”. Sībawayhi explains lam tašġal ʿan-hā qabl-a ʾan talḥaq-a ʾillā al-fiʿl-a bi-ġayr-i-hā “you do not [allow] the verb to be occupied by anything before the insertion of ʾillā” (Kitāb II, p. 311). By depriving a verb of one of its arguments, ʾillā is treated as if it does not exist, and the noun that follows takes the appropriate case ending as dictated by the sentence structure as the final argument of the verb. Therefore, the particle ʾillā does not govern what follows and is treated as if it does not syntactically exist.

However if the mustaṯnā min-hu, the noun before ʾillā, is present, the sentence is complete before reaching the particle ʾillā. In this case, the mustaṯnā or the noun after ʾillā may be considered badal 88. Badal refers to constructions where a noun substitutes for a previous noun and follows it in case. It is generally translated as apposition or substitution. “apposition” standing for the first noun, the mustaṯnā min-hu: the second noun substituting for the first (Kitāb II, p. 311):

It is noteworthy to mention that if ʾaḥad “anyone” is removed from these sentences, the verbs will be missing one argument as mentioned above, and what is left would be sentences (1)–(3) above, where the particle ʾillā does not govern and does not syntactically exist. However, since the mustaṯnā min-hu, the noun before ʾillā, is present, and therefore all the arguments of the verb are present, the appropriate structure, Sībawayhi affirms, is badal. Sībawayhi explains that hāḏā waǧh-u al-kalām-i ʾan taǧʿal-a al-mustaṯnā badal-an min al-laḏī qabl-i-hi liʾann-a-ka tudḫil-u-hu fī-mā ʾaḫraǧta min-hu al-ʾawwal-a “this is the proper utterance. You make the excepted [noun] substitute for [the noun] that precedes [ʾillā.], because you introduce [the excepted noun] to where the first [noun] has been excluded” (Kitāb II, p. 311). He considers this construction the most appropriate and preferred option, because both nouns are equal in status and one can replace the other. He supplies the reader with further examples all illustrating the same badal constructions.

Within badal structures, both the nominative and accusative are acceptable under certain conditions when using sentential verbs, because, in Arabic, they need two accusatives.99.Sentential verbs are verbs that need sentences as their complements. In Arabic, a specific class of verbs called ʾafʿāl al-qulūb “verbs of the heart” are introduced to complete equational sentences with a mubtadaʾ and a ḫabar, subject and predicate noun, and govern both in the accusative case. A simple example will suffice. Both parts of the following equational sentence turn into two accusatives when the verb ẓanna is inserted: al-walad-u ṭālib-un “The boy is a student.” The boy is the mubtadaʾ and student is the ḫabar. Once the verb ẓanna is inserted, such as ẓanan-tu al-walad-a ṭālib-an “I thought that the boy is a student” both mubtadaʾ and ḫabar become accusative as objects of the verb ẓanna “to think.” His examples include (Kitāb II, p. 313):

The accusative case is chosen as a substitution for ʾaḥad “anyone”, and the nominative is chosen as a substitution for the subject pronoun in the verb yaqūl-u “he says”, since both are objects of the verb ʾaẓunn-u “I think.” Consequently, the noun after ʾillā may substitute ʾaḥad “anyone” and therefore is accusative, or it may substitute the subject pronoun in the verb yaqūl-u “he says” and therefore is nominative. Nevertheless, these sentential verbs do not convey actions. They convey what is in the speaker’s mind. Sībawayhi maintains that these verbs are not in the same category as “verbs of action”, wa-ʾinnamā yadull-u ʿalā mā fī ʿilm-i-ka “but they represent what is in your knowledge” i.e., what is in the mind (Kitāb II, p. 314); and therefore can also be called “verbs of the mind”. As for verbs conveying action such as (p. 313):

only the accusative is permissible, because only one object is needed, and the noun after ʾillā can only be in apposition to ʾaḥad “anyone.”

Sībawayhi (Kitāb II, p. 314) also introduces badal in different syntactic positions. The substitution for the position of mubtadaʾ “subject” is illustrated in examples including:

where Zayd does not substitute for raǧul-un “man” but for qalla rajul-un “hardly a man”, functioning as the mubtadaʾ, therefore Zayd receives the nominative case. Substitution for the position of mubtadaʾ is also seen (p. 318) in:

where Zayd also substitutes for the mubtadaʾ lā ʾaḥada “no one” and receives the nominative case.

Sībawayhi (p. 317) also presents substitution with other syntactic positions, such as the noun after ʾinna:

where Zayd substitutes for the accusative noun after ʾanna, which, in this case, has been dropped from the sentence, and therefore Zayd is accusative

Other similar badal constructions occur in sentences (Kitāb II, p. 315) such as:

where in both cases min ʾaḥad-in “of anyone” is equivalent to ʾaḥad-un “anyone” in the nominative, functioning as the subject of ʾatā “to come”, or ʾaḥad-an “anyone” in the accusative, functioning as the object of raʾā “to see” in (12) and (13), respectively. Both cases are badal constructions.1010.The preposition min “of” in both sentences daḫalat hunā tawkīd-an “is inserted here for emphasis”, as explained by Sībawayhi (Kitāb II, p. 316).

As for the alternate choice Sībawayhi gives at the beginning of the chapter, when he mentions that the noun after ʾillā receives the accusative case, he states that the accusative is also acceptable since baʿḍ-u al-ʿarab-i al-mawṯūq-i bi-ʿarabiyyat-i-hi “some Arabs whose Arabic is trustworthy” use the accusative (Kitāb II, p. 319). Sībawayhi cites the same examples used with badal constructions in sentences (4)–(6) above, but instead of alternating between cases depending on the case of the substituted noun, all employ the accusative case (ibid.):

If the nouns after ʾillā are accusative, then they are not badal, but they are considered munqatiʿ “detached”, and the verbs governing the first nouns, i.e., the nouns before ʾillā, the mustaṯnā min-hu ʾaḥad “anyone” do not have an effect on what comes after ʾillā, the mustaṯnā Zayd. It is interesting that Sībawaayhi uses the term munqatiʿ “detached” in these cases, because the meaning is clearly different from what later grammarians used. Later grammarians used istiṯnāʾ munqatiʿ “detached exception” for when the excepted noun is heterogeneous with the munstaṯnā min-hu (Carter 1975Carter, Michael G. 1975 “A Note on Classical Exceptive Sentences”. Journal of Semitic Studies 20.69–72. : 69), i.e., both nouns do not belong to the same category.

Once again, the verbs in examples (14)–(16) have all their arguments, just as they did in badal sentences (4)–(6) above. However, Sībawayhi explains that in structures similar to sentences (14)–(16), where the tanwīn-naṣb structure is used, they are not badal structures. Verbs in these sentences do not govern nouns after ʾillā, unlike what has been stated earlier with sentences (4)–(6) in the badal structures. This is clarified by Sībawayhi’s statement wa-ḏālika ʾanna-ka lam taǧʿal al-ʾāḫir-a badal-an min al-ʾawwal-i wa-lākinna-ka ǧaʿalta-hu munqatiʿ-an mim-mā ʿamil-a fī al-ʾawwal-i “you do not substitute the second [noun] for the first [noun] but you detach it from what governs the first [noun]” (Kitāb II, p. 319). Sībawayhi explains this discrepancy by explaining why these structures are not badal. The meaning of ʾillā, in these specific sentences, occurs as lākin “but” or lā ʾaʿnī “I do not mean”, in effect meaning, “No one came but Zayd”, or “No one came; I do not mean Zayd.” As Baalbaki (2008Baalbaki, Ramzi 2008 The Legacy of the “Kitāb”: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: Brill. : 56–68) explains, similarity in meaning can be an ʿilla “cause” according to Sībawayhi, which he sometimes employs to justify usage. This resemblance in meaning could have triggered Sībawayhi’s explanation of the accusative in his illustration. In both cases Zayd is accusative, in the first as the subject-noun following lākin, and in the second as the object of ʾaʿnī. Sībawayhi asserts that both constructions are acceptable: badal and ʾinqiṭāʿ “apposition and detachment”, although he prefers badal, as stated earlier.

However, badal constructions are not acceptable if the substitution does not make sense, in which case, the noun after ʾillā must occur detached. Such structures occur when the noun after ʾillā cannot be an element of or cannot relate to the first noun.1111.In other words, heterogeneous with the munstaṯnā min-hu. Examples such as (Kitāb II, p. 319):

perfectly illustrate this concept, since a donkey cannot be an appropriate substitute for humans.

Sībawayhi presents more examples upholding the accusative choice. He explains that nouns after ʾillā must be accusative in sentences such as (p. 331):

His justification is that the mustaṯnā is not included in the mustaṯnā min-hu. ʾAb-ā-ka “your father” is not part of al-qawm “the people” who actually came, and, therefore, the verb ʾatā “to come” does not govern ʾab-ā-ka “your father” and is detached. Sībawayhi explains that the reason is liʾanna-hu muḫriǧ-un mim-mā ʾadḫalta fī-hi ġayr-a-hu fa-ʿamila fī-hi mā qabl-a-hu kamā ʿamila al-ʿišr-ūna fī al-diham-i “because [the noun that follows ʾillā] is excluded from what has been included [before]. Therefore ʾillā operates on it just like ʿišr-ūna operates on dirham” in a tanwīn-naṣb construction (Kitāb II, p. 330). He continues intaṣaba al-ʾab-u ʾiḏ lam yakun dāḫil-an fī-mā daḫala fī-hi mā qabl-a-hu wa-lam yakun ṣifa wa-kāna al-ʿāmil-u fī-hi mā qabl-a-hu min al-kalām-i “ʾab ‘father’ is accusative because it is not included in what has been included before [ʾillā]; it is not an attribute, and what precedes operates on it” as in a tanwīn-naṣb construction (ibid., p. 331). He further elaborates by explaining the difference between two types of sentences (ibid.):

and sentence (18) above. Sentence (19) is acceptable with the mustaṯnā ʾab-ū-ka “your father” in the nominative case, because had the mustaṯnā min-hu, al-qawm “the people”, been missing, the resulting sentence (ibid.):

is acceptable since it conveys the same meaning. In this case, the badal structure is appropriate, and thus, the nominative case.

However, if the mustaṯnā min-hu, al-qawm “the people”, is missing from (18), the resulting sentence would be (ibid.):

This is an absurd or an inconceivable utterance, or in Sībawayhi’s words muḥāl. Sentences (18) and (21) do not mean the same thing, and therefore, the structure cannot be badal. The only possible choice is munqaṭiʿ “detached” structure in which the function of the verb ʾatā “he came” does not govern the mustaṯnā ʾab-ā-ka “your father.” The only explanation that potentially clarifies this ambiguity is Sībawayhi’s unique way of describing negated and affirmed sentences. For negated sentences, his preferred structure for the noun after ʾillā, the mustaṯnā, is badal, since the presence of the mustaṯnā min-hu is optional, because, as stated above, the particle ʾillā deactivates the negation of the verb, and thus the verb governs the mustaṯnā as its subject when the mustaṯnā min-hu is not present. Sībawayhi uses vague sentences to explain his argument. He says that the mustaṯnā in negated sentences tudḫil-u-hu fī-mā ʾaḫraǧta min-hu al-awwal-a “is included in what the [mustaṯnā min-hu] has been excluded” (Kitāb II, p. 311), so the verb is allowed to govern it. However, the mustaṯnā in affirmed sentences is not part of the mustaṯnā min-hu, as observed in sentence (18) above. This exclusion makes it impossible for the verb to govern the noun after ʾillā, the mustaṯnā, which must be “detached” from the verb. In affirmed sentences, the mustaṯnā muḫriǧ-un mim-mā ʾadḫalta fī-hi ġayr-a-hu “is excluded from what the [mustaṯnā min-hu] has been included” (ibid., p. 330), so the verb is not allowed to govern it. Essentially, it can only be accusative. With this in mind, it is clear that there is a choice with negated sentences: either badal or detachment, but, for affirmed utterances, the only option is accusative.

An additional mandatory accusative structure is when ʾillā and its noun, the mustaṯnā, are fronted, i.e., occurring before the noun that would otherwise be substituted, the mustaṯnā min-hu. If the mustaṯnā occurs at the beginning of the sentence, it cannot be badal structure, because the noun that is supposed to be replaced does not appear before the mustaṯnā. According to Sībawayhi, and this only applies to negated sentences, when the badal is an option, al-istiṯnāʾ ḥadd-u-hu ʾan tadārak-a-hu baʿd-a-mā tanfī fa-tubdil-u-hu “the definition of exception is that you reach it after negation then you substitute it with [the previous noun]” (Kitāb II, p. 335). Essentially the noun after ʾillā can only occur as the noun that substitutes for a previous noun; otherwise, there is no badal construction. Sībawayhi uses the sentence (ibid.):

with an accusative mustaṯnā, noun after ʾillā, to illustrate this structure, with ṣadīq “friend” occurring after the exceptive phrase, and therefore forcing the accusative case on ʾab “father”, because a badal structure is not allowed. However, Sībawayhi also introduces what seems to be the opposite structure in (p. 336):

with a nominative ʾab “father” after ʾillā. His explanation here is that ʾab “father” is a substitute for man “who” in a badal structure and ṣadīq “friend” is in a ḥāl “circumstantial state” in the accusative.

Sībawayhi introduces a non-exceptive ʾillā, where it occurs as an attribute meaning ġayr “other than.” Ibn Mālik explains that ġayr “other than” primarily occurs as an attribute, and ʾillā “except” primarily occurs in istiṯnāʾ, but then both assume the [role] of the other in where each primarily occurs ( Tashīl Ibn Mālik, Tashīl = Jamāl al-Dīn Muḥammad b. ʿAbdullāh Ibn Mālik, Šarḥ al-Tashīl Ed. by ʿAbdulraḥmān al-Sayyid and Muḥammad Badawī al-Maḫtūn. Vol. II. Cairo: Haǧr li-l-Ṭibāʿa wa-l-Našr wa-l-Tawzīʿ wa-l-ʾIʿlān 1990., pp. 297–298). In this case the noun after ʾillā cannot be in a badal structure, and the whole phrase can only be an attribute (Sībawayhi Kitāb II, p. 331):

Sībawayhi explains that if what is meant were an exceptive phrase, it would be an impossible utterance (p. 331):

He further explains that if you say sentence (26) wa-ʾanta turīd-u al-istiṯnāʾ la-kunta qad ʾaḥalta “and you want the exception, you would have said an absurd [utterance]” (p. 331). Sībawayhi does not explain the impossibility of this utterance, but Ibn Mālik relates Sīrāfī’s explanation that a badal structure in istiṯnāʾ necessitates an affirmed, or non-negated mustaṯnā, noun after ʾillā ( Tashīl Ibn Mālik, Tashīl = Jamāl al-Dīn Muḥammad b. ʿAbdullāh Ibn Mālik, Šarḥ al-Tashīl Ed. by ʿAbdulraḥmān al-Sayyid and Muḥammad Badawī al-Maḫtūn. Vol. II. Cairo: Haǧr li-l-Ṭibāʿa wa-l-Našr wa-l-Tawzīʿ wa-l-ʾIʿlān 1990., p. 299), in which case it would give a different meaning indicating that if Zayd were present, they would have been defeated, and this is not what is meant by sentence (25). Effectively, Sīrāfī gives the sentence law kāna maʿa-nā Zayd-un la-halak-nā “Had Zayd been with us, we would have perished” (Ibn Mālik Tashīl Ibn Mālik, Tashīl = Jamāl al-Dīn Muḥammad b. ʿAbdullāh Ibn Mālik, Šarḥ al-Tashīl Ed. by ʿAbdulraḥmān al-Sayyid and Muḥammad Badawī al-Maḫtūn. Vol. II. Cairo: Haǧr li-l-Ṭibāʿa wa-l-Našr wa-l-Tawzīʿ wa-l-ʾIʿlān 1990., p. 299). Additionally, he tells us that al-badal-u baʿd-a ʾillā fī al-istiṯnāʾ-i mūjab-un “badal after ʾillā is affirmative in exception” (ibid.), since a badal can only occur in a negated sentence in exception. So, if sentence (25) has a badal structure, Zayd would not be excluded, and this proves the impossibility of the utterance. On the contrary, it indicates that without Zayd, they would have been defeated.

Sībawayhi cites one Quranic verse (21:22) that uses this non-exceptive ʾillā (Kitāb II, p. 332):

If Allah 1212.I use the conventionally written Allah in the text as opposed to the transliteration I use in the example: al-lāh. “God” were a badal, Allah “God” would substitute for ʾālihah “gods” and ultimately put “God” in place of “gods”; and, therefore, would translate according to Sīrāfī as related by Ibn Mālik ( Tashīl Ibn Mālik, Tashīl = Jamāl al-Dīn Muḥammad b. ʿAbdullāh Ibn Mālik, Šarḥ al-Tashīl Ed. by ʿAbdulraḥmān al-Sayyid and Muḥammad Badawī al-Maḫtūn. Vol. II. Cairo: Haǧr li-l-Ṭibāʿa wa-l-Našr wa-l-Tawzīʿ wa-l-ʾIʿlān 1990., p. 299) as:

This is an unthinkable expression in the Quran. However, this sentence means (Ibn al-ʾAnbārī ʾInṣāf, p. 235):

Treating ʾillā as the attribute ġayr would change the meaning of the sentence to a perfectly acceptable expression.

Sībawayhi never explains the nominative “God” after ʾillā, except by treating it as an attribute not a badal. He also never explains the function of the conditional law “if”. Later grammarians explain further by reiterating that badal only occurs in negated sentences. Ibn al-ʾAnbārī states that al-badal-u fī al-ʾiṯbāt-i ġayr-u ǧāʾiz-in “substitution in affirmed sentences is not permissible” (ʾInṣāf, p. 235). Astarābāḏī states that al-badal-u lā yaǧūz-u ʾillā fī ġayr-i al-mūǧab-i “substitution is only allowed in non-affirmed sentences” (Šarḥ al-Raḍī, p. 185). Zamaḫšarī states that al-badal-u lā yasūġ-u ʾillā fī al-kalām-i ġayr-i al-mūǧab-i “substitution is not allowed except in non-affirmed sentences” (Kaššāf, p. 86). Others add that conditionals indicate affirmed sentences. Ibn Yaʿīš states that al-šarṭ-u fī ḥukm-i al-mūǧab-i “conditionals are virtually affirmed [sentences]” (Mufaṣṣal, p. 89), and Ibn Mālik indicates that speech is affirmative with conditionals ( Tashīl Ibn Mālik, Tashīl = Jamāl al-Dīn Muḥammad b. ʿAbdullāh Ibn Mālik, Šarḥ al-Tashīl Ed. by ʿAbdulraḥmān al-Sayyid and Muḥammad Badawī al-Maḫtūn. Vol. II. Cairo: Haǧr li-l-Ṭibāʿa wa-l-Našr wa-l-Tawzīʿ wa-l-ʾIʿlān 1990., p. 298). Astarābāḏī states that laysa al-šarṭ-u…min ġayr-i al-mūǧab-i “conditionals are not beyond affirmed [sentences]” (Šarḥ al-Raḍī, p. 185). Finally, Zamaḫšarī indicates that speech with law “if” is affirmative (Kaššāf, p. 86). Furthermore, Ibn Yaʿīš asserts that the function of ʾillā in this verse is to indicate the antithesis (Mufaṣṣal, p. 89). He further elaborates that [ʾillā] wa-mā baʿd-a-hā taḥliyat-un li-l-maḏkūr-i bi-l-muġāyarat-i “[ʾillā] and [the noun] that follows describe the attributes of what was mentioned [before] as contradictory” (Mufaṣṣal, p. 89). He continues explaining that the case of the noun after ʾillā follows the case of the noun before ʾillā, which explains the nominative case in “God.” In explaining the nominative, Ibn Yaʿīš states that mā baʿd-a ʾillā fī al-waṣf-i yakūn-u ʾiʿrāb-u-hu tābiʿ-an li-ʾiʿrāb-i mā qabl-a-hā “[the noun] after ʾillā [occurring] as an attribute follows the previous noun in case” (Mufaṣṣal, pp. 89–90). It is not a badal for “gods”, but an attribute in antithesis, thus receiving the same case ending as “gods.”

This is what Baalbaki (1995Baalbaki, Ramzi 1995 “Reclassification in Arab Grammatical Theory”. Journal of Near Eastern Studies 54:1.1–13. : 7) calls “Reclassification according to grammatical function”, where grammarians allowed a shift from the ʾaṣl to explain certain constructions, which otherwise would have been unacceptable. In this case, ʾillā as a particle has been modified to function as the noun ġayr, as an attribute, in order to explain the nominative after the particle ʾillā, which otherwise would have been badal, indicating an unacceptable utterance in the Quran. The noun after ʾillā in affirmed sentences is accusative, though, in order to maintain this “basic rule,” this construction in this verse is considered something other than istiṯnāʾ. The shift in grammatical category is necessary to explain this anomaly. Sībawayhi “opts to confirm an exception to the boundaries between the parts of speech rather than an exception to the specific case under discussion,” namely the istiṯnāʾ (Baalbaki 2008Baalbaki, Ramzi 2008 The Legacy of the “Kitāb”: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: Brill. : 164, 166).

In brief, Sībawayhi starts his chapters on ʾillā with the ʾillā of no effect in negated sentences. He then presents the badal as his primary choice if the sentence is complete before ʾillā, revealing other badal possibilities, especially those that involve different syntactic positions. What is remarkable is his non-technical use of the word “detached” to explain the tanwīn-naṣb construction if the sentence is complete before ʾillā. Since the verb cannot operate on the word after ʾillā, it is considered detached, and the tanwīn-naṣb structure is allowed. He also uses the word “detached” when referring to different categories, as will be used by later grammarians as a technical term in the development of the grammatical theory.

3.Zaǧǧāǧī’s Ǧumal

The topic of istiṯnāʾ in Zaǧǧāǧī’s Ǧumal is presented with clear rules without considerable analysis, and examples serve as proof as well as good illustrative models for the rules. Unlike Sībawayhi, Zaǧǧāǧī starts with affirmative sentences (Ğumal, p. 235):

In which case, the mustaṯnā must be in the accusative case in the tanwīn-naṣb structure. However, if the sentence is negated, you have a choice (ibid.):

The mustaṯnā is nominative under badal rules or accusative if there is a complete sentence before the particle ʾillā.

These are simple rules. If the sentence is affirmed, then the noun after ʾillā is accusative, but if it is negated, then the noun may be badal or accusative in a tanwīn-naṣb structure, but only if there is a complete sentence before the particle ʾillā. Zaǧǧāǧī states that ʾiḏā kāna mā qabl-a [ʾillā] min al-kalām-i mūǧab-an kāna mā baʿd-a-hā manṣūb-an … “if what comes before ʾillā is affirmed in utterance, then what comes after it is accusative” (Ğumal, p. 235). He continues wa-ʾiḏā kāna ma qabl-a ʾillā ġayr-a mūǧab-in kāna mā baʿd-a-hā tābiʿ-an li-mā qabl-a-hā ʿalā al-badal-i wa-ǧāza fī-hi al-naṣb-u ʾiḏā tamma al-kalām-u dūn-a-hu “and if what comes before ʾillā is not affirmative then what comes after it modifies what comes before in substitution, and the accusative is permissible if the utterance is complete before [the ʾillā phrase]” (ibid.).

However, if the mustaṯnā min-hu is not present, the verb governs what follows. He elaborates ʾiḏā farraġta mā qabl-a ʾillā li-mā baʿd-a-hā ʿamila fī-hi wa-lam taʿmal ʾillā šayʾ-an “if you free [the verb] before ʾillā for what follows, then [the verb] governs [what follows] and ʾillā does not govern anything” (Zaǧǧāǧī Ğumal, p. 236). This is the istiṯnāʾ mufarraġ “exhaustive exception” as used by later grammarians. The use of the verb farraġa “to free”, I believe, does not yet convey the technical term, but it indicates the opposite of šaġala “to operate”, because Zaǧǧāǧī already uses the technical terms istiṯnāʾ muqaddam “fronted exception” (ibid., p. 238) and istiṯnāʾ munqatiʿ “detached exception” (p. 239). If istiṯnāʾ mufarraġ “exhaustive exception” was already considered a technical term at this time, he would have used it in the same way he used the other technical terms. Zaǧǧāǧī gives the following examples if you remove the mustaṯnā min-hu (Ğumal, p. 236):

Zayd in sentence (30) is nominative because it is the subject of the verb, Zayd in sentence (31) is accusative because it is the object of the verb, and Zayd in sentence (32) is genitive because it is the object of a preposition. This applies to sentences where the verb does not have all its arguments before the exception particle ʾillā and thus has no effect on what follows.

Zaǧǧāǧī continues with other examples where the accusative case is the only choice: if the mustaṯnā is fronted, or if the mustaṯnā and the mustaṯnā min-hu do not belong to the same category, where both mustaṯnā and mustaṯnā min-hu cannot replace one another. Zaǧǧāǧī explains that ʾiḏā kāna al-mustaṯnā min ġayr-i ǧins-i al-ʾawwal-i kāna munqaṭiʿ-an min-hu manṣūb-an “if the noun after ʾillā is not of the same category as the first [noun], then it is detached and is accusative” (Ğumal, p. 239). Zaǧǧāǧī gives the following examples if the noun after ʾillā is fronted (p. 238):

Similarly, if we have two different categories (p. 239):

The mustaṯnā must be accusative, because ḥimār “donkey” is not in the same category as ʾaḥad “anyone”. We here have two species, and thus, one cannot substitute for the other.

Zaǧǧāǧī contrasts affirmed and negated sentences at the beginning of his chapter, where tanwīn-naṣb structure is the only choice or is preferred to badal in negated sentences, but only if the sentence is complete before ʾillā; otherwise, ʾillā has no effect. Zaǧǧāǧ’s primary concern is the difference between affirmed and negated sentences, and based on this distinction rests his grammatical choice.

4.Ibn Bābašāḏ’s Muqaddima

As for the Muqaddima of Ibn Bābašāḏ, istiṯnāʾ is presented with direct, straightforward explanations providing only one example for each rule. According to Ibn Bābašāḏ, exception indicates ʾiḫrāǧ-u baʿḍ-in min kull-in bi-ʾillā “extracting some from [the] whole with [the use] of ʾillā.” (Ibn Bābašāḏ Naḥwiyya (Šarīf) II, p. 263). Exception is used to extract something from all else. He explains that the noun after ʾillā is accusative if it is in an affirmed sentence or fī taʾwīl-i al-mūǧab-i “what is similar to an affirmed sentence” (ibid., p. 264):

which is similar to an affirmed sentence meaning that everyone but Zayd ate bread.

The mustaṯnā is also accusative if it precedes the mustaṯnā min-hu or is munqatiʿ “detached”, as respectively expressed by the following sentences (ibid.):

The latter must be detached because it deals with two different species.

The accusative for the mustaṯnā is also a possibility baʿd-a tamām-i al-kalām-i “after a complete sentence” (ibid.):

Ibn Bābašāḏ also explains in his commentary that the mustaṯnā in sentence (40) can also be considered badal, and would thus receive the nominative case, only because the mustaṯnā occurs after a complete sentence (ibid., p. 269):

where the noun after ʾillā qalīl “few” would substitute for the subject pronoun in the verb faʿal-ū-hu “they did it.” Ibn Bābašāḏ does not give the alternative example where qalīl “few” is nominative in his Muqaddima, but he elucidates in his commentary in both editions (ibid., p. 269; Ibn Bābašāḏ Muḥsiba (Abd al-Karīm) II, p. 324).

Ibn Bābašāḏ further states that if the sentence starts with an interrogative, it is negated, or prohibitive, the mustaṯnā, ġālib-an “usually”, follows the mustaṯnā min-hu in case (Ibn Bābašāḏ Naḥwiyya (Šarīf) II, p. 264), i.e., a badal structure:

While this example is given as part of the original Muqaddima as an example for nafy “negation”, in his commentary, edited by both Šarīf and ʿAbd al-Karīm, a slightly different version is presented (ibid. p. 270; Ibn Bābašāḏ Muḥsiba (ʿAbd al-Karīm) II, p. 325):

While the translation is not affected, however, the difference between the negation particles lam and mā is noteworthy, which leads me to think that Ibn Bābašāḏ did in fact chose a different example in his commentary. This is possible since the Muqaddima and its commentary were written almost thirty years apart.

For the nahy “prohibitive” (Ibn Bābašāḏ Muqaddima, manuscript, leaf 22; Ibn Bābašāḏ Muḥsiba (ʿAbd al-Karīm) II, p. 321), he gives:

This example occurs in the manuscript as well as in ʿAbd al-Karīm’s edition. In the Šarīf edition, however, Ibn Bābašāḏ only provides this example for nahy “prohibitive” in the commentary (Naḥwiyya (Šarīf) II, p. 270). Given that the manuscript in my hand, ʿAbd al-Karīm’s edition, and the commentary edited by Šarīf have the same example, I speculate that that the manuscripts Šarīf used are to blame or that it is an editorial slip on the part of Šarīf.

His use of ġālib-an “usually” reflects the fact that the accusative is also allowed if the utterance is complete before ʾillā, i.e., if the mustaṯnā min-hu is present and thus serves its function in the sentence and completes all the arguments of the verb, as in sentence (40) above.

Ibn Bābašāḏ presents all tanwīn-naṣb “accusative” possibilities, whether affirmed sentences, fronted, detached, or after a complete sentence if negated, at the beginning of his chapter. He then mentions the badal only as a possibility with negated sentences. Ibn Bābašāḏ’s primary concern is to emphasize the tanwīn-naṣb structure as primary and the badal only as a possibility.

5.Concluding observations

Comparing these istiṯnāʾ “exception” excerpts, the fundamental rules are the same; however, presentation varies widely, particularly when comparing Sībawayhi with the latter grammarians. Rules become simpler as one progresses in time, and lengthy explanations come to an end. In addition to the lack of extensive justification in the latter two selections, different badal “apposition” structures, elaborated by Sībawayhi are missing from Zaǧǧāǧī and Ibn Bābašāḏ. Whereas Sībawayhi’s treatment of badal “apposition” in istiṯnāʾ “exception” is primary, since he deals with negated structures first, it is only mentioned as a possibility in Zaǧǧāǧī and Ibn Bābašāḏ, as their primary istiṯnāʾ “exception” structure is the tanwīn-naṣb structure: accusative. Furthermore, Zaǧǧāǧī and Ibn Bābašāḏ already use the term “detached” as a technical term referring to nouns that belong to two different categories. The word, as used by Sībawayhi, has dropped from usage as the other two grammarians found no reason to justify the tanwīn-naṣb “accusative” structure as Sībawayhi did. Likewise, the controversial attributive ʾillā is missing from both Zaǧǧāǧī’s Ǧumal and Ibn Bābašāḏ’s Muqaddima. This is possibly because it is not considered an exceptive, a structure clearly not required to be mastered by beginners. This similarity between Zaǧǧāǧī and Ibn Bābašāḏ may also originate in the fact that the latter composed a commentary on Zaǧǧāǧī’s Ǧumal. Ibn Bābašāḏ further neglects mentioning istiṯnāʾ “exception” when the mustaṯnā min-hu “noun before ʾillā” is missing, as ʾillā in this case is neutralized.

Although neither Zaǧǧāǧī nor Ibn Bābašāḏ, justify usage in presenting their rules, the main difference between their presentations is the number of examples used. Zaǧǧāǧī has seventeen examples in addition to three poetic references and four Quranic verses, while Ibn Bābašāḏ only uses seven examples (eight in the manuscript and the ʿAbd al-Karīm edition) in addition to two Quranic verses. This is in complete contrast to Sībawayhi who uses more than 120 examples in addition to nineteen poetic references and seven Quranic verses. Although Zaǧǧāǧī’s Ǧumal and Ibn Bābašāḏ’s Muqaddima are brief and concise and evidently should be used for beginners, other grammarians found the need to write extensive commentaries expanding and clarifying missing justifications. That grammar needed to be speculative was clearly expected by scholars of the period. These commentaries, however, serve as necessary substitutes for instructors, who were essential in the teaching of these manuals and often offered their own commentaries along with their teaching.

It is clear that Sībawayhi describes the language in order to generate rules of usage by examining multitudes of examples taken from the language in use. His explanations justify the rules he deduces from the data. He is therefore obligated to present these lengthy clarifications to rationalize his rules. Rules of istiṯnāʾ emerge from within his explanations, yet he further explains any structures where the rules do not exactly correspond. By the time of Zaǧǧāǧī these explanations were stripped from all but the most necessary rules for the exception. He gives clear explanations but still finds the need occasionally to supply multiple examples to support these rules, albeit not close in number to Sībawayhi’s. However, Ibn Bābašāḏ finds that he needs only one example to convey each rule. Although the last two books do include minimal speculation and justification, they are definitely more accessible to learners for their clarity and simple presentation. Lengthy justifications are not needed for learners of the language. They have their acceptable, and well appreciated, role in speculative and theoretical grammars.