‘Five’ as a typological number in the medieval Arabic grammatical tradition

In the Arabic grammatical tradition several categories comprising exactly five members can be found, e.g., the types of “meaningful things”, of definite nouns, of tanwīn, of definite article, of tawābiʿ etc. Given the importance of the number ‘five’ in Islam, it is natural to ask whether these categorizations are affected by the symbolical meaning of that number. This article examines some of these categorizations in order to check the extent to which they are linguistically or theoretically justified, and whether they use ‘five’ as a typological number. In order to answer these questions, the fivefold divisions are tested for consistency and the surrounding discourse is investigated.


Introduction
Organizing material into short lists may be a powerful mnemonic and pedagogical tool, but in light of the frequent use of the specific number 'five' in Arabic grammatical literature, we ask whether this particular number has a special meaning and what that meaning can be.
Various cultures ascribe symbolic and even magical significance to numbers. Numerical symbolism in monotheistic religions is probably related to the Pythagorean tradition, which holds that the cosmic order can be expressed by numbers. This tradition views odd numbers in general as auspicious (and even numbers as boding ill). In cultures influenced by this tradition, ritual acts and prayers are repeated an odd number of times. 1 Among various meaningful numbers, the number five is related to the pentagonal symmetry, to the five senses, etc. (although, unlike three and seven, five is not considered mysterious). 2 • 19 (2019):  ing-in-a-position". 20 According to al-Jāḥiẓ, this modality of clarity refers to phenomena of the natural world that can be interpreted, for instance, as signs of God's presence. 21 Montgomery claims that there was "an intellectual and polemical connection between the legaltheoretical Risāla ('Epistle') of Šāfiʿī and the Bayān of Jāḥiẓ" 22 (note that al-Šāfiʿī's Risāla also presents five wujūh 'modalities' of bayān, 23 and so does al-Jaṣṣāṣ' Fuṣūl 24 ).
In the grammatical literature, traces of these ideas can be found in discussions of the definition of 'word' by Ibn Yaʿīš (d. 1245) andal-ʾAstarābāḏī (d. circa 1289).
Ibn al-Ḥājib (d. 1249) defines 'word' as lafẓ wuḍiʿa li-maʿnan mufradin 'an expression that was coined for a simple meaning' 28 (a definition that is very close to al-Zamaḫšarī's, which is to be expected, given the proven affinity between the latter's al-Mufaṣṣal and Ibn al-Ḥājib's al-Muqaddima al-Kāfiya 29 ). Al-ʾAstarābāḏī explains this definition, and says that mentioning lafẓ is necessary because it excludes from the definition al-ḫaṭṭ, al-ʿaqd, alniṣba and al-ʾišāra, which may signify a simple meaning by their coinage, but cannot be considered words. 24 Al-Jaṣṣāṣ (Fuṣūl II: 14-19) discusses al-Šāfiʿī's classification of bayān types. He also mentions (ibid.: 31) a classification similar to al-Jāḥiẓ's, attributing it to "one ancient scholar". See BERNAND 1995: 152 for a discussion.
25 Coinage (waḍʿ) is an important term in Muslim philosophy, theology and jurisprudence. In the grammatical literature, it is particularly prominent in al-ʾAstarābāḏī's Šarḥ al-Kāfiya. Al-ʾAstarābāḏī (Šarḥ I: 21) explains the phrase "the coinage of a linguistic expression" (waḍʿ al-lafẓ) as "the first assignment of [a linguistic expression] to a meaning, with an intention that it will become conventional between people." An element's coinage determines its form, meaning, categorical identity, syntactic functions, etc.
(although in actual use there can be certain deviations from those primary properties of the element). Al-ʾAstarābāḏī's theory of coinage is discussed in SHEYHATOVITCH 2018: 84-127. 26 Ibn Yaʿīš, Šarḥ I: 18. See GUILLAUME 2011: 51-53 for a discussion of al-Zamaḫšarī's definition (with a stress on its Aristotelian origin).
27 Ibn Yaʿīš, Šarḥ I: 19. in Kitāb al-Lumaʿ uses the term ʾasmāʾ al-ʾišāra instead of mubhamāt 36 ). The order of categories (and, correspondingly, the hierarchy of definite nouns) varies from one author to another. 37 Al-Zamaḫšarī seems to be the first grammarian who explicitly included the relative pronouns in the category of mubhamāt, together with demonstratives, in the context of discussing definite nouns. His list of categories is: "proper nouns" (al-ʿalam al-ḫāṣṣ), "personal pronouns" (al-muḍmar), "vague nouns, that include two things: demonstrative pronouns and relative pronouns" (al-mubham wa-huwa šayʾāni ʾasmāʾu l-ʾišārati wa-l-mawṣūlātu), "[nouns] prefixed by the definite article" (al-dāḫil ʿalayhi ḥarfu l-taʿrīfi), "nouns annexed in a real annexation to one of these [types of nouns]" (al-muḍāf ʾilā ʾaḥadi hāʾulāʾi ʾiḍāfatan ḥaqīqiyyatan). 38 Ibn Yaʿīš explains that the principal difference between mubham and muḍmar, vague and pronominalized constituents, is that a vague constituent is elucidated by a constituent that follows it, whereas a 3 rd person pronoun is elucidated by a constituent that precedes it, i.e., by its antecedent. 39 Ibn ʿUṣfūr (d. 1271) presents the following five categories of definite nouns: personal pronouns, demonstrative pronouns, proper nouns, nouns prefixed by the definite article, nouns annexed to definite nouns in a real annexation. He includes relative pronouns in the category of nouns preceded by the definite article, stating: fa-ʾammā l-mawṣūlātu fa-min qabīli mā ʿurrifa bi-l-ʾalifi wa-l-lāmi, but mentions a controversy related to the question of whether they are made definite by an overt article or an intended one. 40 Ibn ʿUṣfūr presents two different opinions: according to ʾAbū ʿAlī l-Fārisī (d. 987), relative pronouns "are definite by the previous knowledge (i.e., by virtue of information that is known equally to the speaker and the addressee), represented by the relative clause" (taʿarrafat bi-l-ʿahdi llaḏī fī l-ṣilati). According to ʾAbū l- , relative pronouns are definite due to the definite article. The former opinion is supported by the fact that some relative pronouns come without the definite article (for instance, mā and man). The latter is supported by the claim that definiteness can exist only with the definite article or annexation, while the case of mā and man is explained by analogy to saḥar 'this dawn': if the latter refers to the dawn of the day on which the utterance is produced, it is considered as definite (and thus is diptote), but comes without the definite article, due to ʿadl 'anomaly'. 41 The view that allaḏī etc. are definite due to the definite article may be challenged by the claim that some relative pronouns can create annexation structures, but the answer would be that when functioning as an annexed 36 Ibn Jinnī,Lumaʿ,159. 37 See GÄTJE 1970, GABUČAN 1972: 40-41, and MAROGY 2010  element, the relative pronoun is definite due to the annexation, and its original definit eness is ignored. 42 Al-ʾAstarābāḏī, similarly to Ibn Yaʿīš, regards relative pronouns, together with demonstrative pronouns, as mubhamāt, 43 but explains the reason for their definiteness differently: he does not consider them as definite due to the definite article (he says that their definiteness resembles that of nouns preceded by al-, 44 implying that it is not the same). Also, he does not accept the view that the relative pronoun is definite due to the relative clause: he argues that if the relative clause could render other constituents definite, it should also have rendered definite the noun rajul in the sentence jāʾanī rajulun ḍarabtuhu 'A man whom I hit came to me". 45 So what is al-ʾAstarābāḏī's explanation for the relative pronoun's definiteness?
The definiteness of the relative pronoun is due to its coinage as a definite noun that is used to refer to a thing known to the speaker and the addressee, [whereas the common knowledge regarding that thing is] the content of the relative clause [that follows that pronoun] (taʿrīfu l-mawṣūli bi-waḍʿihi maʿrifatan mušāran bihi ʾilā l-maʿhūdi bayna l-mutakallimi wa-l-muḫāṭabi bi-maḍmūni ṣilatihi). 46 In other words, the definiteness of allaḏī should be explained by the hypothetical coiner's intention rather than by a formal factor.
To sum up, the above-mentioned grammarians accept the fivefold division of definite nouns, although some of them obviously struggle to find a right place for relative pronouns within their categories. Hence it is not surprising that some grammarians challenged the accepted categorization and treated relative pronouns as a separate category.
Ibn Mālik (d. 1274) presents the definite nouns as follows: (1) It may mark the distinction between mutamakkin (lit. 'powerful; established', fully declinable) 51 and light constituents, on the one hand, and heavy constituents that are not fully declinable, on the other hand; 52 (2) It may serve as compensation (ʿiwaḍ) for an omitted part of the word-for example, the form jawārin 'female slaves' is derived from a supposed original form *jawāriyu in rafʿ or *jawāriyi in jarr; the sequences -iyu and -iyi are considered "heavy", and therefore yāʾ becomes quiescent; then the pattern of the word becomes deficient, and the word is joined by a compensating tanwīn, becoming *jawāriyn, and then the yāʾ is omitted, because it is a quiescent letter followed by a vowelless nūn. This is the process by which the form jawārin is obtained. 53 Al-Zajjājī uses this as a central example in his discussion on "a compensating tanwīn", as its tanwīn is clearly unrelated to tamakkun because the word is diptote by its pattern. (3) It may make it possible to distinguish between definite and indefinite states in proper nouns and interjections whose ending is usually left unchanged. For instance, the proper noun ʿAmrawayhi retains its ending as long as it is used as semantically definite, but once it becomes semantically indefinite (namely, when it is used to signify a class of objects 54 ), the proper noun takes a tanwīn to signify that it is used as an indefinite noun. For instance, hāḏā ʿamrawayhi wa-marartu bi-ʿamrawayhin ʾāḫara 'This is ʿAmrawayhi and I passed by another [man named] ʿAmrawayhi'. 55 49 See GABUČAN 1972: 55-56 for a discussion of al-ʾUšmūnī's (d. 1464) view of tanwīn as an indefinite article, and of modern linguists who adopted this view. AYOUB (1991: 169 ff.) examines this view, and concludes that it is insufficient as an explanation of all appearances of tanwīn, given that proper nouns such as Zayd take tanwīn despite their semantic definiteness. She offers a more complicated explanation for the tanwīn function: it is related to the nominality of the noun in different modules (the module of logical form, the module of cases and the syntactic module). See ibid.: 207-209 for a discussion. 50 The following excerpt is translated and discussed also in VERSTEEGH 1995: 168-176. 51 Mutamakkin is an active participle derived from the term tamakkun, which, in DANECKI's (2009: 431) words, "is used for a general grammatical and semantic category indicating the ability of words to be inflected and perform various grammatical functions". Al-Zamaḫšarī is probably the first grammarian to speak of five types of tanwīn. They are the following:

wa-ġayruhu maʿrifatun ka-hum wa-ḏī / wa-hinda wa-bnī wa-l-ġulāmi wa-llaḏī
(1) Al-dāll ʿalā l-makāna 'signifying the status'. According to Ibn Yaʿīš, this type of tanwīn signifies that the noun retains its nominal status, becoming neither like a particle (which would render the noun mabnī) nor like a verb (which would render the noun diptote). 56 This type of tanwīn parallels the first type mentioned by al-Zajjājī. Unlike al-Zajjājī, who concentrates on the morphophonological aspects of the phenomenon and speaks of "compensating tanwīn" in relation to words whose third root consonant is wāw or yāʾ, al-Zamaḫšarī and Ibn Yaʿīš speak here about forms such as yawmaʾiḏin 'that day', where, according to Ibn Yaʿīš, the tanwīn compensates for an omitted clause (or clauses). For instance, Q 99:1-4: ʾiḏā zulzilat-i l-ʾarḍu zilzālahā wa-ʾaḫrajat-i l-ʾarḍu ʾaṯqālahā wa-qāla l-ʾinsānu mā lahā yawmaʾiḏin tuḥaddiṯu ʾaḫbārahā 'When earth is shaken with a mighty shaking and earth brings forth her burdens, and Man says, "What ails her?", upon that day she shall tell her tidings'. The reconstructed original structure is: …yawmaʾiḏin tuzalzalu l-ʾarḍu zilzālahā watuḫriju l-ʾarḍu ʾaṯqālahā wa-yaqūlu l-ʾinsānu mā lahā… 'Upon the day when earth is shaken with a mighty shaking and brings forth her burdens, and Man says, "What ails her?", [she shall tell her tidings]'.
The three clauses that should have functioned as a governed element of the first word were omitted, and the tanwīn took their place. The form should have been *yawmaʾiḏn, and a kasra was added after the ḏāl to prevent a sequence of two vowelless consonants. 58 (4) Al-nāʾib manāb ḥarf al-ʾiṭlāq "[tanwīn] that takes the place of the long vowel of a loose rhyme". 59 Traditional Arabic poetic theory recognizes two types of rhyme (qāfiya), muqayyada 'fettered' and muṭlaqa 'loose'. In the former, the rhyme consonant is not followed by a letter of prolongation. In the latter, a letter of prolongation is attached. Various types of rhyme where the rhyme consonant is followed by a short vowel and a vowelled or quiescent hāʾ are also called 'loose rhyme'. 60 So alclassifying the tanwīns of ʿamrawayhin and of ʾibrāhīmin as belonging to two different categories is somewhat problematic-see ibid.: 445). It can be concluded that tanwīn as a marker of indefinite nouns (widely known as tanwīn al-tankīr) is irrelevant for originally diptote proper nouns that receive tanwīn due to their semantic indefiniteness. That is because once ʾibrāhīm ceases to be semantically definite, it ceases to behave as a diptote proper noun, behaving instead as a regular triptote noun. Consequently, it is appropriate for it to take tanwīn al-tamakkun. Zamaḫšarī speaks here of a tanwīn used instead of a vowel prolongation at the end of poetry verses, a practice characteristic of Banū Tamīm's recitation. (5) Al-tanwīn al-ġālī 'extravagant tanwīn 61 ' that joins only a 'fettered' rhyme. 62 Interestingly, al-Zamaḫšarī presents tanwīn as one of the "exclusive attributes" (ḫaṣāʾiṣ) of a noun 63 (which does not prevent him from distinguishing five types of tanwīn, including those peculiar to poetry, that can also join verbs). Ibn Yaʿīš notices this theoretical difficulty, and resolves it by explaining that when speaking of a tanwīn as an exclusive attribute of a noun, al-Zamaḫšarī has in mind tanwīn al-tamkīn only (called al-dāll ʿalā l-makāna by al-Zamaḫšarī), which indeed is attached only to nouns, in order to distinguish between those that are fully declinable and those that are not. Ibn Yaʿīš says that al-Zamaḫšarī could not have meant "an absolute tanwīn" (muṭlaq al-tanwīn, i.e., tanwīn in general), as this category includes also tanwīn al-tarannum that can also join verbs. 64 After explaining the first three types of tanwīn mentioned by al-Zamaḫšarī, Ibn Yaʿīš states that the fourth type is tanwīn al-tarannum, which appears only in poetry and is related to rhyme 65 (in Ayoub's words, it is used "so as to produce a musical effect" 66 ). Nūn can replace letters of prolongation, which it resembles, being a nasal consonant. Tanwīn altarannum can, in turn, be divided into two types: a tanwīn that makes the verse's structure and pattern complete, and a tanwīn that is added after all parts of the verse are present. The former is added in a loose rhyme, the latter in a fettered rhyme. The two types of tanwīn altarannum correspond to al-Zamaḫšarī's fourth and fifth types. Ibn Yaʿīš criticizes al-Zamaḫšarī for classifying those as two separate types of tanwīn, and also for omitting another type-tanwīn al-muqābala 'tanwīn of comparison'. This tanwīn is added to proper nouns that have the form of the sound feminine plural. For instance, Muslimāt, when functioning as a feminine proper noun, is expected to be diptote (like any feminine proper noun); however, it takes a tanwīn, to match the nūn of the ending of sound masculine plural forms, such as muslimūna. 67 To sum up, Ibn Yaʿīš recognizes the following five types of tanwīn: li-l-farq bayna mā yanṣarifu wa-mā lā yanṣarifu, al-dāll ʿalā l-nakira, tanwīn al-ʿiwaḍ, tanwīn al-tarannum, tanwīn al-muqābala. 68 It seems that he chose to join the two types of rhyme-related tanwīns under the title tanwīn al-tarannum in order to maintain the fivefold categorization.
Ibn ʿUṣfūr presents the same five types: tanwīn li-l-tamakkun, tanwīn al-tankīr, tanwīn al-muqābala, tanwīn al-ʿiwaḍ (unlike al-Zamaḫšarī and Ibn Yaʿīš, he mentions in this context both cases like yawmaʾiḏin and cases like jawārin), tanwīn al-tarannum (he mentions only the case of loose rhymes, i.e., the first of the two cases mentioned by Ibn Yaʿīš); he stresses that all types except for the last-mentioned one join only nouns. 69 Al-ʾAstarābāḏī deals with tanwīn in a chapter dedicated to "the exclusive attributes of a noun" (ḫawāṣṣ al-ism). 70 He mentions the same five types of tanwīn, distinguishing from the outset between tanwīn al-tarannum, that can join various constituents, and the other four types, that are added to nouns only. 71 Unlike the grammarians we mentioned before, who limit tanwīn al-tankīr to certain proper nouns and interjections, al-ʾAstarābāḏī says that the ending of the proper nouns in rubba ʾaḥmadin wa-ʾibrāhīmin 'many men named ʾAḥmad and ʾIbrāhīm' is not only the marker of indefiniteness, but also a marker of establishment, 72 which is the case also with the ending of rajulun 'a man', since there are particles/morphemes (ḥarf) that perform two functions simultaneously. If rajulun functions as a proper noun, its tanwīn should be construed as a marker of establishment only. 73 Al-ʾAstarābāḏī's discussion of tanwīn al-muqābala is relatively detailed and complicated. He does not limit this type of tanwīn to proper nouns, but notes that the fact that it is retained in proper nouns, e.g., in Q 2:198 min ʿarafātin 'from ʿArafāt', proves that it is not a marker of establishment (since proper nouns that are feminine by form or meaning should be diptote) nor of indefiniteness (because proper nouns are definite). 74 However, his final conclusion is that it is "[a marker] of full declension and establishment" (li-l-ṣarf 75 wa-ltamakkun). At this point he needs to explain why this tanwīn is not omitted in min ʿarafātin. His explanation is as follows: If [the tanwīn] was omitted, it would have been followed by the kasra in the omission, and the naṣb would have followed [the omitted kasra]. That would have differed from the usual state of affairs in the sound plural, where the kasra is followed by others, instead of following others (li-ʾannahu law saqaṭa la-tabiʿahu l-kasru fī lsuqūṭi, wa-tabiʿa l-naṣbu, wa-huwa ḫilāfu mā ʿalayhi l-jamʿu l-sālimu ʾiḏ-i l-kasru fīhi matbūʿun lā tābiʿun). 76 In other words, if ʿarafāt had been a diptote noun, it would not have taken kasra, as a consequence of its not receiving a tanwīn. 77 Thus, its ending in jarr would have become identical to its ending in naṣb, which should not happen in the sound plural, in which naṣb generally behaves analogously to jarr, but not vice versa. 78 According to al-ʾAstarābāḏī, the tanwīn in min ʿarafātin resembles a tanwīn in a diptote noun that received it "due to poetic license" (li-l-ḍarūra), as there was something that prevented the tanwīn's omission. 79 It should be mentioned that al-ʾAstarābāḏī considers tanwīn, regardless of its specific function, a marker of a word's end and an indicator that the word is not an annexed element (dāllan ʿalā tamāmi l-kalimati wa-ʾannahā ġayru muḍāfatin). This trait is common to the tanwīn and the nūn of the suffix of the dual and sound masculine plural, but the above-mentioned five functions pertain to tanwīn only. 80 Despite the problematic status of tanwīn al-tarannum (which can join nouns and verbs, and thus seems to contradict the claim that tanwīn is an exclusive attribute of a noun), and despite blurring the difference between tanwīn al-muqābala and tanwīn altamakkun, al-ʾAstarābāḏī keeps speaking of five types of tanwīn.
Al-Širbīnī (d. 1570) views tanwīn as one of the identifying features of a noun. He focuses on the four types of tanwīn that are added only to nouns: tanwīn al-tamakkun (he mentions that it is called also tanwīn al-ʾamkaniyya and tanwīn al-ṣarf 81 ), tanwīn altankīr (which, like most grammarians, he limits to baʿḍ al-ʾasmāʾ al-mabniyyāt 'certain invariable nouns'), 82 tanwīn al-muqābala (which he does not limit to proper nouns), tanwīn al-ʿiwaḍ. As for the latter type, al-Širbīnī says that it joins nouns such as ġawāšin 'covers' and jawārin "in compensation for the arbitrarily omitted yāʾ" (ʿiwaḍan min-a l-yāʾi l-maḥḏūfati ʿtibāṭan), and also ʾiḏ in cases such as Q 30:4 yawmaʾiḏin yafraḥu l-muʾminūna 'on that day the believers shall rejoice, where the tanwīn compensates for the omitted governed element (which should be a clause). Al-Širbīnī cites Ibn Hišām (d. 1360), who argues that the tanwīn in kullun 'all' and baʿḍun 'some' is also tanwīn al-ʿiwaḍ that compensates for an omitted governed element. Al-Širbīnī rejects this claim, arguing that this is the tanwīn of establishment "which disappears in an annexation and remains in the absence [of annexation]" (yaḏhabu maʿa l-ʾiḍāfati wa-yaṯbutu maʿa ʿadamihā). 83 Al-Širbīnī notes that some grammarians add another six types of tanwīn to this list. He cites an anonymous verse that includes the entire list of ten. The additional types are: (1) redundant, (2) 1991: 208) notes that in this al-ʾAstarābāḏī differed from other grammarians, who were unable to relate the distinct values of tanwīn to each other and to a general property of a category of nouns. She offers a more nuanced analysis: in her view, tanwīn is a marker of syntactic completeness when suffixed to a noun that functions as an argument, whereas it is a marker of syntactic incompleteness when suffixed to a noun that functions as a predicate. See AYOUB 1991: 198 ff. for a discussion. 81 CARTER 1981: 16-17 (CARTER translates these three terms, respectively, as "the tanwīn of establishment", "the tanwīn of stability", and "the tanwīn of currency"). cense, (5) metrical extravagance, (6) after hamza. 84 After that he states that these types "are called tanwīn in a non-literal sense, not in a literal one, since they are not peculiar to nouns" (tasmiyatuhā tanwīnan majāzun lā ḥaqīqatun li-ʿadami ḫtiṣāṣihā bi-l-ismi). 85 This is how al-Širbīnī reconciles between his view of tanwīn as one of the identifying features of a noun and the fact that some sources mention tanwīns that can also join other parts of speech.

Five types of definite article
Ibn ʿUṣfūr follows his presentation of tanwīn with a categorization of functions of the definite article: (1) li-taʿrīfi l-ʿahdi fī šaḫṣin ʾaw fī jinsin 'for a definiteness [based on the addressee's] previous knowledge, of an individual or a genus' (usually, grammarians view taʿrīf al-ʿahd and taʿrīf al-jins as different categories, 86 but Ibn ʿUṣfūr probably thinks that speakers may have previous knowledge of genera just like they may have previous knowledge of individuals, and thus these two functions of the definite article are closer to each other than to its other functions). The examples are jāʾanī l-rajulu llaḏī jāʾaka 'The man that came to you came to me' (the referent of the noun prefixed by the definite article is an individual known to both the speaker and the addressee), and al-rajulu ḫayrun min-a l-marʾati 'Man is better than woman' (this genus is better than that genus). 87 (2) li-taʿrīfi l-ḥuḍūri 'for a definiteness based on presence'. Unlike taʿrīf al-ʿahd, which is a definiteness based on previous knowledge, this definiteness is related to the referent's being present at the moment of utterance. This is the definite article that joins the noun that follows a demonstrative pronoun, e.g., hāḏa l-rajulu 'this man', or a vocative particle, e.g., yā ʾayyuhā l-rajulu 'O the man!', or ʾiḏā l-fujāʾiyya, 88 e.g., ḫarajtu fa-ʾiḏā l-ʾasadu 'I went out and there was the lion'. This is also the definite article that appears in words such as al-ʾān, al-sāʿa, al-ḥīn 'now'. 89 (3) li-lamḥi l-ṣifati 'because of an intimation of the adjective'. This is a definite article that is added to a proper noun that was originally an adjective, such as al-Ḥāriṯ, lit. 'the plowing one' and al-ʿAbbās lit. 'the frowning one'. This alis not an insepara- ble part of these words, as can be proven by the fact that one may say rajulun ḥāriṯun 'a plowing man' and rajulun ʿabbāsun 'a frowning man'. 90 (4) li-l-ġalaba 'for a predominance'. This is a definite article that joins an indefinite noun in order to render it definite, and then dominates it. The noun consequently becomes a proper noun. For instance, al-Najm, lit. 'the star', which denotes the Pleiades; if the intended meaning is the Pleiades (and not just any star), the alis obligatory. 91 (5) zāʾida 'redundant'. This type of article is attached to proper nouns that do not originate in adjectives, and is used only due to poetic license. For instance: ʾa-mā wa-dimāʾin lā tazālu murāqatan / ʿalā qunnati l-ʿuzzā wa-bi-l-nasri ʿandamā 'I swear by blood that still penetrates / the upper part of al-ʿUzzā [idol] and renders al-Nasr blood-red' 92 The poet used al-Nasr instead of Nasr (lit. 'vulture'; the name of an idol worshipped by the Ḥimyarites 93 ).
After mentioning these types of al-(of which there were five), Ibn ʿUṣfūr adds: "These four types are found only in nouns" (wa-hāḏihi al-ʾaḍrubu l-ʾarbaʿatu lā tūjadu ʾillā fī l-ʾasmāʾi ḫāṣṣatan), and then starts a discussion on alwith the meaning of allaḏī, which joins participles, but in poetry can also join verbs and clauses. 94 His mention of "four types" is truly surprising, because he clearly listed five types just before. Perhaps this is due to an attempt on his part to create a symmetry between the categorization of tanwīns (five types, of which one type, tanwīn al-tarannum, can join various parts of speech, and the rest can join only nouns), and the categorization of al-(again five types, of which one type can sometimes be attached to verbs and clauses, and the rest only to nouns). In defense of Ibn ʿUṣfūr's logic, it can only be said that the types of althat join only nouns could be viewed as four instead of five if we ignore the "redundant" al-(which has no meaning or function besides regularizing the rhythm of a poetic verse). It is the only type in Ibn ʿUṣfūr's categorization which is mentioned without the preposition li-'for', and this may mean that the author did not intend to include it in the total. Ibn Hišām's classification of alis completely different (which proves that Ibn ʿUṣfūr's classification is by no means necessary). The former dedicates a chapter of his book to al-, which he divides into three types: (1) A relative pronoun with the meaning of allaḏī and its likes, which usually joins participles. It can also join time/place expressions and nominal or verbal clauses (if the predicate is an imperfect verb The definite article in al-rasūl signifies that the word's referent is the same as of the previously-mentioned rasūl. Gabučan notes that this description corresponds to the European notion of "anaphoric article". 99 (ii) "[The referent of the noun related to it] is known, because it is present in the [addressee's] mind" (maʿhūdan ḏihniyyan), e.g., Q 9:40 ʾiḏ humā fī l-ġāri 'when the two were in the Cave'. 100 Interestingly, this is the first time "the cave" is mentioned in the sūrah; Ibn Hišām considers its referent as "present in the [addressee's] mind", because the Muslim tradition links it to a specific event when Muḥammad and ʾAbū Bakr were hiding in a cave near Mecca, and their enemies were not able to find them. 101 (iii) "[The referent of the noun related to it] is known, because it is present [in the speech situation]" (maʿhūdan ḥuḍūriyyan]. Here Ibn Hišām cites Ibn ʿUṣfūr's description of cases that the latter classifies as li-taʿrīf alḥuḍūr (the second item in Ibn ʿUṣfūr's classification mentioned above). This proves that Ibn Hišām was acquainted with Ibn ʿUṣfūr's work, and supports an assumption that his classification of alis a reformulation of the latter's.
Ibn Hišām criticizes some points in Ibn ʿUṣfūr's description. First, when the sentence lā taštum-i l-rajula 'Do not curse the man!' is produced by a speaker who witnesses someone cursing a third party, the word alrajul is definite because its referent is present in the speech situation (a case which Ibn ʿUṣfūr did not mention ʾiḏā l-fujāʾiyya "does not signify the definiteness of a thing present at the speech situation, nor a definiteness that resembles [a definiteness of a thing] present at the speech situation" (laysa li-taʿrīfi šayʾin ḥāḍirin ḥālata l-takallumi fa-lā tušbihu mā l-kalāmu fīhi). 102 This criticism is based on the fact that the thing mentioned after ʾiḏā l-fujāʾiyya is present at the situation presented in the sentence (which may be different from the speech situation 103 ). However, Ibn ʿUṣfūr does not mention the speech situation in his discussion, mentioning only the idea of presence (which is relevant in the case of ʾiḏā l-fujāʾiyya)-thus, Ibn Hišām's criticism does not seem justified. Finally, Ibn Hišām says that the alin al-ʾān should be viewed as redundant (zāʾida) rather than as a definite article. That is because this alis an inseparable part of the word, whereas the definite article is not known to appear as an inseparable part of the word. 104 (b) jinsiyya: 105 (i) "Signifying [that the word] comprises all the individuals [that can be referred by it]" (li-stiġrāqi l-ʾafrādi). This is the al-"that can be replaced by kull in its literal sense" (allatī taḫlufuhā kullun ḥaqīqatan). For instance, Q 4:28 wa-ḫuliqa l-ʾinsānu ḍaʿīfan 'for man was created a weakling'. 106 According to the principle formulated by Ibn Hišām, al-ʾinsān is definite in a generic, inclusive definiteness, which can be demonstrated by replacing it by the phrase kullu ʾinsānin in its literal sense. (ii) "Signifying [that the word's referent] comprises all the exclusive attributes of individuals [that can be referred to by the word]" (li-stiġrāqi ḫaṣāʾiṣi al-ʾafrādi). This is the al-"that can be replaced by kull in its non-literal sense" (allatī taḫlufuhā kullun majāzan). For instance, zaydun-i l-rajulu ʿilman 'Zayd is the man in terms of knowledge'. 107 This sentence can be paraphrased by zaydun kullu rajulin ʿilman 'Zayd equals all men in terms of knowledge', i.e., Zayd's knowledge is perfect. 108 • 19 (2019): 81-111 (iii) "That renders the essence definite" (li-taʿrīfi l-māhiyyati). This alcannot be replaced by kull, in either its literal or its non-literal sense. For instance, if someone says wa-llāhi lā ʾatazawwaju l-nisāʾa 'By God, I will not marry the women', his vow would be violated if he married even a single woman. 109 (3) zāʾida 'redundant': (a) "[that constitutes] an inseparable part [of the noun]" (lāzima). That is the alof relative pronouns, according to the view whereby they are rendered definite by the relative clause, 110 and of proper nouns, "provided that [al-] joined them when they were transmitted (from their original meaning to functioning as proper nouns)" (bi-šarṭi muqāranatihā li-naqlihā 111 ), which is the case with al-Naṣr, al-Naʿmān, al-Lāt, al-ʿUzzā, or "when they were invented" (lirtijālihā), which is the case with al-Samawʾal, or when "the word became dominant [in referring to] one of its original referents" (li-ġalabatihā ʿalā baʿḍi man hiya lahu fī l-ʾaṣli), which is the case with al-Bayt, lit. 'The House' (referring to the Kaʿba), al-Madīna, lit. 'The City' (referring to al-Ṭayyiba 'The Good', the city in Saudi Arabia), and al-Najm, lit. 'The Star' (referring to the Pleiades). 112 This last case of "redundant inseparable al-" corresponds to the fourth item in Ibn ʿUṣfūr's classification. 113

(b) "[that does not constitute] an inseparable part of the word" (ġayr lāzima).
This category is divided into two types: (i) A common type, that appears in eloquent speech. This is the althat appears in proper nouns "that were transferred from [common nouns] without [al-] that can be prefixed by it, is such a way that [a proper noun includes] an intimation [of the original common noun]" (manqūl min mujarrad ṣāliḥ lahā malmūḥ ʾaṣluhu). This is the case, e.g., of al-Ḥāriṯ, lit. 'the plowing one', al-ʿAbbās 'the frowning one', and al-Ḍaḥḥāk 'the one who laughs frequently'. 114 (ii) An uncommon type of al-, which may appear in poetry and in anomalous instances of prose. 115 • 19 (2019): 81-111

The five types of tawābiʿ
Tawābiʿ (sing. tābiʿ) is a category of words "whose case assignment is caused by their 'following' another word". This category created a special problem for the Arab grammarians, as its case assignment cannot be explained by governance. 116 Sībawayhi does not speak of such a category. Al-Mubarrad in his treatise uses various terms related to this category, without however systematizing them. 117 Ibn al-Sarrāj is probably the first one to speak of tawābiʿ as a category, that comprises five types of sentence constituents: (1) Tawkīd 'emphasizer'. This type is divided into two: (a) "emphasis by repetition of the noun" (tawkīd bi-takrīr al-ism): (i) "A type in which a noun is repeated literally" (ḍarb yuʿādu fīhi lismu bi-lafẓihi). Despite this formulation, Ibn al-Sarrāj demonstrates that constituents that are repeated can be nouns, verbs, particles (together with the nouns that receive jarr from them) and even clauses. 118 Perhaps he uses the term ism here, although he has in mind various types of constituents, because of the principle that any word can be viewed as a noun, if it is produced with the linguistic expression itself in mind (rather than its meaning/referent). 119 (ii) "[A type in which] the meaning is repeated in different words" (ʾiʿādat al-maʿnā bi-lafẓin ʾāḫara). For instance, marartu bi-zaydin nafsihi 'I passed by Zayd himself'. 120 (b) "signifying comprehensiveness and generality" (li-l-ʾiḥāṭa wa-l-ʿumūm). Ibn al-Sarrāj has in mind the derivatives of words such as kull, ʾajmaʿ, ʾaktaʿ 'all', and also kilā/kiltā 'both (masc./fem.)', ṯalāṯatuhum 'the three of them', etc. 121 116 See the discussions of tawābiʿ in CARTER 1981: 148-149, 238 ff. (where the term is translated as "concordants"), OWENS 1988: 57-58, 162 ff. (where the term is translated as "modifiers"), VERSTEEGH 2009: 221. As for the governor of tawābiʿ, Ibn al-ʾAnbārī (ʾAsrār, 294-295) presents two views regarding the governor of an adjectival qualifier: Sībawayhi's view, that its governor is the same as the head's, and ʾAbū Ḥasan al-ʾAḫfaš's, that the qualifier's rafʿ is assigned by its being a tābiʿ of a head in rafʿ etc. Ibn al-ʾAnbārī notes that the former view is more widely accepted (see LEVIN 1995: 215 for a discussion of additional sources that mention the latter view).
117 For instance, while discussing the structure yā naṣru naṣrun naṣran he uses the terms badal, bayān (which probably corresponds to ʿaṭf bayān), and maʿṭūf (see al-Mubarrad, Muqtaḍab IV: 210-211); while discussing the structure yā hāḏā l-ṭawīlu 'O this tall one!' he explains that al-ṭawīl is ʿaṭf bayān rather than naʿt (see ibid.: 220). (2) Naʿt 'adjectival qualifier'. Interestingly, Ibn al-Sarrāj distinguishes five types of it (and thus creates a fivefold division inside another fivefold division, proving his interest in this number): (a) "An attribute [related to the appearance 122 ] of the described thing, which may exist in [that thing] or in something related to it logically" (mā kāna ḥilyatan li-l-mawṣūfi takūnu fīhi ʾaw fī šayʾin min sababihi). 123 The examples for the first option are: marartu bi-rajulin ʾazraqa/ʾaḥmara/ṭawīlin/qaṣīrin 'I passed by a blue/red/tall/short man'; the examples for the second option are: marartu bi-rajulin ḥasanin ʾabūhu 'I passed by a man whose father is good' and maḍaytu ʾilā rajulin ṭawīlin ʾaḫūhu 'I went to a man whose brother is tall'. 124 (b) "An action of the described thing, which may be performed by [the latter] or by something related to it logically" (mā kāna fiʿlan li-l-mawṣūfi yakūnu bihi fāʿilan ʾaw muttaṣilan bi-šayʾin min sababihi). The examples for the former option are: marartu bi-rajulin qāʾimin/nāʾimin/ḍāribin 'I passed by a standing/sleeping/hitting man'; the examples for the latter option are marartu birajulin ḍāribin ʾabūhu 'I passed by a man whose father is hitting', marartu bi-rajulin qāʾimin ʾaḫūhu 'I passed by a man whose brother is standing', raʾaytu rajulan ḍāriban ʾaḫūhu ʿamran "I saw a man whose brother is hitting ʿAmr', etc. 125 (c) "An attribute [of the described thing], that is neither [its] action nor is related to its appearance" (mā kāna ṣifatan ġayra ʿamalin wa-taḥliyatin), e.g., marartu bi-rajulin ʿālimin/ʿāqilin 'I passed by a knowledgeable/reasonable man', marartu bi-rajulin ʿālimin ʾabūhu 'I passed by a man whose father is knowledgeable', marartu bi-rajulin ẓarīfatin jāriyatuhu 'I passed by a man whose female slave is charming'. (d) "Relation" (nasab). These are adjectives that relate a person or thing to a father, a place, a profession, or some category. For instance marartu bi-rajulin hāšimiyyin/ʿarabiyyin 'I passed by a Hāšimī/Bedouin man' (with adjectives that relate the man to certain genera), marartu bi-rajulin bazzāzin/ʿaṭṭārin/ najjārin "I passed by a man who is a seller of cloth/a seller of perfumes/a carpenter' (with adjectives that relate the man to things with which he deals), 122 This addition is based on one of the definitions of ḥilya given in LANE (1865 II: 635): "the appearance in respect of colour, or complexion, &c., of a man". The addition seems necessary, in order to highlight the difference between this type of qualifier and type (c), that includes attributes which have no external manifestation (see below). 123 See the discussion of the term sabab in CARTER 2009. 124 Ibn al-Sarrāj, ʾUṣūl I: 409-410. Such constructions are known as naʿt sababī, an adjective that refers to the main noun "in virtue of a following word which is connected with it". See WRIGHT 1967II: 283-284. DIEM (1998 renders naʿt sababī as "adjektivischer Satz" since, on the one hand, it is equivalent to a sentence, and on the other hand it has the same syntactic distribution as adjectives and participles. See DIEM 1998 for a detailed discussion on the uses of naʿt sababī. • 19 (2019): 81-111 marartu bi-rajulin baṣriyyin/miṣriyyin/kūfiyyin 'I passed by a Baṣran/Egyptian/Kūfan man' (with adjectives that relate the man to places). 126 (e) "Describing [things] by means of ḏū 'possessor'" (al-waṣf bi-ḏī). For instance, marartu bi-rajulin ḏī ʾiblin 'I passed by a man who possesses camels', marartu bi-rajulin ḏī ʾadabin 'I passed by a man who possesses manners', marartu bi-rajulin ḏī ʿaqlin 'I passed by a man who possesses reason'. 127 This classification seems somewhat artificial, especially the distinction between (a) and (c), and may serve as yet another proof of Ibn al-Sarrāj's adherence to the number 'five' (which is also apparent from the examples surveyed by Okazaki). 128 (3) ʿAṭf al-bayān 'the explicative'. Ibn al-Sarrāj explains that the difference between this constituent and an adjectival qualifier is that the former is an underived noun. The grammarians avoid calling it an adjectival qualifier, because it does not signify an attribute of the described thing and does not correspond to any type of qualifier. It is called 'the explicative' because it distinguishes between the referent of its head and other objects with the same name. For instance: raʾaytu zaydan ʾabā ʿamrin 'I saw Zayd, ʿAmr's father' and laqītu ʾaḫāka bakran 'I met your brother Bakr'. 129 The difference between ʿaṭf al-bayān and apposition is that the former is intended to be similar to an adjectival qualifier of the head, and the latter is intended to be a constituent that can replace the head. Thus, when the speaker uses ʿaṭf al-bayān, he says yā ʾaḫānā zaydan 'O our brother Zayd!' 130 (zaydan takes naṣb, according to the basic rule regarding words in vocative 131 ), and when he uses apposition, he says yā ʾaḫānā zaydu (zaydu takes ḍamma, as a single noun that follows a vocative particle, as if it followed yā directly). 132 131 According to Ibn al-Sarrāj, the vocative particle yā governs the following noun phrase similarly to the verb ʾunādī 'I call'. However, a definite single noun takes the bināʾ ending ḍamma, because such a noun occurs in a position characteristic of personal pronouns (since it denotes a second person instead of its regular meaning). See the discussion in ibid.: 300.
• 19 (2019): 81-111 (5) Al-ʿaṭf bi-l-ḥarf 'coordination by means of a particle'. Ibn al-Sarrāj states that there are ten coordinative particles that "make the constituent that follows them to agree [in case] with the noun or verb that precedes them" (yutbiʿna mā baʿdahunna mā qablahunna min-a l-ʾasmāʾi wa-l-ʾafʿāli). 134 Ibn Jinnī mentions the same five types of tawābiʿ; 135 Ibn al-ʾAnbārī does not speak explicitly of tawābiʿ as a category, but places in a succession the chapters dealing with tawkīd, waṣf (a term very close to naʿt), ʿaṭf al-bayān, badal, and ʿaṭf. 136 Al-Zajjājī appears to be the first grammarian who speaks about four tawābiʿ: al-naʿt, al-ʿaṭf, al-tawkīd, al-badal (he omits ʿaṭf al-bayān). 137 Ibn ʿUṣfūr does not challenge this categorization, but incorporates ʿaṭf al-bayān by distinguishing between two types of ʿaṭf: ʿaṭf al-bayān and ʿaṭf al-nasq 'coordination'. 138 Al-Baṭalyawsī (d. 1127) openly criticizes al-Zajjājī's categorization of tawābiʿ: he says that al-Zajjājī did not mention ʿaṭf al-bayān, as if it can be included in his four categories, which is not true. Al-Baṭalyawsī notes that it exists only in definite nouns, and shares some positions with the adjectival qualifier, some with apposition, and some are unique to it (which last point proves that al-Zajjājī's categorization is not accurate). 139 He mentions three positions that are unique to ʿaṭf albayān: 1. In vocative constructions: ʾinnī wa-ʾasṭārin suṭirna saṭran / la-qāʾilun yā naṣru naṣrun naṣrā 'I swear by the lines [of the Qurʾān] that are indeed written, / I say: O Naṣr, Naṣr, Naṣr…' 140 Al-Baṭalyawsī explains that those who assign naṣb to the second and third naṣr, construe them as ʿaṭf al-bayān that agrees with the position of the first naṣr, which follows a vocative particle 141 (because the position following a vocative particle should in principle be occupied by a noun in naṣb; however, a semantically definite noun that is not an annexed element takes a bināʾ ending ḍamma in this position 142 ). assign rafʿ+tanwīn to the second naṣr, construe it as ʿaṭf al-bayān that agrees with the form of the first one (which is possible, because the ḍamma in the ending of a noun that follows a vocative particle resembles an ʾiʿrāb ending 143 ), and construe the third naṣr as ʿaṭf al-bayān that agrees with the position of the first one. Those who assign the second naṣr rafʿ without a tanwīn, construe it as an apposition of the first. 144 As we shall see below, al-Baṭalyawsī holds that in apposition there is the intention of repeating the governor, whereas there is no such intention in ʿaṭf al-bayān. Thus, one can imagine the second naṣr as immediately following a suppressed vocative particle, and consequently assign it ḍamma, provided one construes the second naṣr as an apposition. 2. With vague constituents (i.e., demonstrative pronouns): e.g., marartu bi-hāḏā l-rajuli 'I passed by this man', laqītu hāḏā l-ġulāma 'I met this servant'. Al-Baṭalyawsī notes that grammarians refer to nouns following the demonstrative as naʿt 'adjectival qualifier', whereas it is actually ʿaṭf bayān. 3. With active participles: e.g., hāḏā l-ḍāribu l-rajuli zaydin 'This is the one hitting the man, Zayd'. Zayd can be assigned jarr only if it is construed as ʿaṭf al-bayān of alrajul. In order to be construed as an apposition, it should be a word that can occupy the position of its head, whereas one cannot say *hāḏā l-ḍāribu zaydin 'This is the one hitting Zayd' (instead one should say hāḏā l-ḍāribu zaydan, i.e., use a direct object instead of an annexation structure), because a noun prefixed by the definite article cannot be annexed to a noun that is not prefixed by that article, unless the former noun is in dual or masculine sound plural form. 145 After that, al-Baṭalyawsī takes pains to distinguish between ʿaṭf al-bayān and other tawābiʿ that can be confused with it. He mentions five points of difference between ʿaṭf al-bayān and adjectival qualifiers: 1. The function of a qualifier is usually performed by adjectives, whereas ʿaṭf al-bayān is an underived noun, similarly to apposition. 146 2. A qualifier may be either definite or indefinite, whereas ʿaṭf al-bayān, according to the Baṣran view, must be definite. 147 • 19 (2019): 81-111

l-ʾāna lam yaẓhar lī farqun jaliyyun bayna badali l-kulli min-a l-kulli wa-ʿaṭfi lbayāni, bal lā ʾarā ʿaṭfa l-bayāni ʾillā l-badala). 153
Al-ʾAstarābāḏī argues that Sībawayhi held the same view when he said, with regard to the sentence marartu bi-rajulin ʿabdi llāhi 'I passed by a man, ʿAbdallāh', that there is a substitution of a definite noun for an indefinite one, as if the speaker was asked 'Who did you pass by?', or thought that someone might ask such a question, and thus replaced the indefinite noun with a more definite one. 154 It should be mentioned that al-ʾAstarābāḏī equates ʿaṭf al-bayān with a full substitution, one of the apposition types-therefore, points (1) and (4) from al-Baṭalyawsī's discussion on the difference between ʿaṭf al-bayān and apposition, in which various types of the latter are mentioned, should be irrelevant for him.
Al-ʾAstarābāḏī says that other grammarians may claim that the difference between ʿaṭf al-bayān and apposition is that an apposition (and not its head) is "the one intended in the ascription" (al-maqṣūd bi-l-nisba), whereas ʿaṭf al-bayān is an explanation, and an explanation is secondary in relation to what is explained. Therefore, in the case of ʿaṭf al-bayān the one intended in the ascription is the head. 155 The term nisba in Šarḥ al-Kāfiya refers to semantic relations between constituents, which may be predicative or not. 156 When al-ʾAstarābāḏī speaks of "being intended in the ascription", he probably means that the constituent in question is more important than another one or other ones, it is the one that is really meant to create syntactic connections with other parts of the sentence. 157 The grammarians' claim that the apposition is more important than its head (whereas in the case of adjectival qualifier the head is more important) is well known. 158 Al-ʾAstarābāḏī does not accept the claim that in the case of an apposition the speaker intends the second constituent only; in his view, this is true only for 'permutative of error', in which "the second constituent (i.e., the apposition) is obviously intended instead of the first (i.e., the head)" (fa-ʾinna kawna l-ṯānī fīhi huwa l-maqṣūdu dūna l-ʾawwali ẓāhirun). 159 He 153 Al-ʾAstarābāḏī, Šarḥ II: 379. See SARTORI, forthcoming: Section 1 for an alternative translation of this passage.
154 Al-ʾAstarābāḏī, Šarḥ II: 379. Here al-ʾAstarābāḏī refers to a discussion from Sībawayhi, Kitāb I: 192. However, Sībawayhi did use the term ʿaṭf al-bayān once-while discussing the above-mentioned verse by Ruʾba (see Kitāb I: 263). See TALMON 1981: 282 for a discussion of a relevant excerpt. TALMON (1981) reconstructs the category of "appositival ʿaṭf" in Sībawayhi's al-Kitāb, considering it an important syntactic innovation that can be attributed without doubt to Sībawayhi and al-Ḫalīl, in contrast to grammarians of Sībawayhi's time and of previous generations. According to Talmon's description, Sībawayhi's "appositival ʿaṭf" roughly corresponds to both tawkīd and ʿaṭf bayān in later grammatical literature. 158 For instance, al-Jurjānī (Muqtaṣid II: 930) says: "The apposition behaves as if the governor was repeated before it, because the head is neglected for the sake of the apposition" (wa-ʾinnamā kāna l-badalu fī ḥukmi takrīri l-ʿāmili li-ʾajli ʾanna l-badala yutraku ʾilayhi l-mubdalu minhu). See SARTORI, forthcoming for a discussion on this and similar excerpts.
• 19 (2019): 81-111 ʾAstarābāḏī says in response that "a constituent that is called ʿaṭf al-bayān" (al-musammā ʿaṭfa bayānin; a formulation that stresses again his disapproval of this term) can also disagree with its head in definiteness. 169 Sartori, after analyzing various grammarians' remarks on the difference between the ʿaṭf al-bayān and apposition, reaches the conclusion that the most essential difference is of a suprasegmental character, namely, apposition is preceded in speech by a pause, whereas ʿaṭf albayān is pronounced immediately after the main noun, similarly to an adjectival qualifier. 170 It is unclear whether al-ʾAstarābāḏī did not grasp this essential difference between ʿaṭf al-bayān and apposition, or did grasp it but considered it not significant enough to categorize these two as separate types of tawābiʿ. He repeats his statement that ʿaṭf al-bayān is actually an apposition several times, 171 but, this notwithstanding, constantly speaks of five tawābiʿ. 172 That can be explained either by retaining the accepted views and terminology, or by recognizing the importance of the number 'five' (the former possibility seems less convincing, given al-ʾAstarābāḏī's general non-conformism 173 ).
Interestingly enough, the tendency towards a division into four (instead of five) types of tawābiʿ, that started with al-Zajjājī and Ibn Mālik, became dominant in modern grammatical literature in Arabic. For instance, Ḥasan in his al-Naḥw al-Wāfī dedicates a detailed chapter to "the four tawābiʿ", in which ʿaṭf al-bayān is subsumed under the category of ʿaṭf. 174

Conclusion
This article analyzed several fivefold divisions found in medieval Arabic grammatical literature. It strove to determine to what extent these divisions are theoretically justified and to what extent 'five' appears in them as a typological number (given the special place of this number in Islam).
The first categorization that was discussed was "five types of meaningful things". It was demonstrated that Ibn Yaʿīš and al-ʾAstarābāḏī adopted al-Jaḥiẓ's fivefold categorization, although the distinction between oral speech and writing (and the resulting exclusion of written words from the definition of 'word') seems unjustified in the context of their respective discussions.
As for the fivefold classification of definite nouns, it was accepted by most grammarians starting with Sībawayhi, despite the fact that in such a scheme it is difficult to find the right place for relative pronouns (that may appear as nouns prefixed by a definite article, but differ from them in some features). This difficulty led Ibn ʿAqīl and Ibn Hišām to place