1This paper is inspired by Kouloughli’s 2007 edition of al‑Zamaḫšarī’s al‑Unmūḏaǧ fī al‑naḥw, which makes accessible, by its text, translation and notes, the essence of Arabic syntax, and thus unites the two linguistic traditions of the Arab and the Western worlds.

2One notion for which al‑Zamaḫšarī is well known — indeed notorious in some quarters — is the assertion that the particle lan signifies negation in perpetuity, taʾbīd al‑nafy, the so‑called lan al‑zamaḫšariyya. This heretical proposition enraged the orthodox to a point where, as this paper will argue, it was introduced retrospectively into the manuscript tradition even though it is extremely unlikely that al‑Zamaḫšarī himself ever made such a claim for lan.

3Before going any further, it is necessary to point out that for a number of reasons, particularly of time pressure and inaccessibility of material, the contents of this paper rely on only a fraction of the relevant primary sources, mainly al‑Zamaḫšarī’s Kaššāf, Mufaṣṣal and Unmūḏaǧ, and the commentary of al‑Ardabīlī (see further below) on the Unmūḏaǧ. A valuable resource, which quotes a much wider range of sources, is the online article by Ibrāhīm b. Sulaymān al‑Buʿaymī entitled Qaḍāyā lan fī al‑naḥw al‑ʿarabī.

4The other commentaries on the Unmūḏaǧ were simply out of reach, geographically and electronically. Brockelmann (GAL 1:291 and S 1:510f) lists eight (not including commentaries on the šawāhid), to which a ninth can be added, al‑Fawāʾid al‑ʿabdiyya fī šarḥ Unmūḏaǧ al‑Zamaḫšarī, by al‑Mūstarī (1061/1651‑1151/1738‑1739), reported by al‑Buʿaymī (23, details in n. 158). The anonymous item listed in no. 3 by Brockelmann as MS Garrett 334 is evidently a fragment of the commentary of al‑Bardaʿī.

5Some of these are too late to be of much use, but three are potentially critical, and the conclusions of this paper could be overturned by any one of them. They are, in reverse chronological order:

• Brockelmann no. 2. Ḥadāʼiq al‑ḥaqāʼiq by Saʿd al‑Dīn Saʿd Allāh al‑Bardaʿī (d. ‎927/1520‑1521‎).
• Brockelmann no. 8. An untitled commentary by ʿAlī b. ʿAbd Allāh b. Aḥmad Zayn al‑ʿArab, composed in 736/1335.
• Brockelmann no. 5. Kifāyat al‑naḥw by Ḍiyāʾ al‑Dīn al‑Makkī, who died in 568/1172. This author appears in three separate entries in Brockelmann, with minor differences in the name and biographical details, viz. (1) in S 1,513 as Ḍiyāʾ al‑Dīn al‑Makkī, active 550/1155, (2) in S 1,549 as Ḍiyāʾ al‑Dīn Abū al‑Muʾayyad al‑Muwaffaq b. Aḥmad b. Isḥāq al‑Makkī al‑Bakrī al‑Ḫawārazmi, 484‑568/1091‑1172, and (3) in S 1,623 as Abū al‑Muʾayyad Muwaffaq b. Aḥmad b. Abī Saʿīd Isḥāq al‑Ḫawārazmī, died 568/1172, and there are also variations in his name in the two published editions of his commentary.

7As far as the Unmūḏaǧ is concerned, therefore, the texts on which this paper is based are :

• three versions of the Unmūḏaǧ itself, namely the editions of Sāmī b. Ḥamad al‑Manṣūr (1420/1999) and of Kouloughli (2007) and a downloaded Saudi MS (see bibliography for details); the hand-written edition of Broch (1863) was inaccessible.
• For the commentary of al‑Ardabīlī four versions are used, namely the partial edition of Silvestre de Sacy (1829), and three downloaded Saudi MSS (details in bibliography). Two recent printed editions (see bibliography), one used by al‑Buʿaymī, could not be accessed; however, the three MSS consulted and de Sacy all agree on the relevant passages.

8Little is known about Ǧamāl al‑Dīn Muḥammad b. ʿAbd al‑Ġanī al‑Ardabīlī, and his dates are disputed (see discussion in al‑Buʿāymī, 23 and n. 156). At one time he was believed to have lived in the 11th/17th century, but the existence of at least two MSS from the 9th/15th century pushes the date back. The current view is that he died in 647/1249, which would make him an older contemporary of Ibn Mālik (d. 672/1274, but see al‑Buʿaymī for objections). For reasons which could not be ascertained the editor of the version used by al‑Buʿaymī (Šāḏilī Farhūd) gives the author’s name as al‑Arbilī, presumably after the Kurdish town of Arbil, while all other sources call him al‑Ardabīlī after Ardabīl in Azerbaijan.

9In this paper the focus will be restricted to the perpetual negation assigned to lan, and other aspects will be left out of account. These include the etymology of lan and the related issue of whether lan is a dependence operator in its own right or because of the ʾan element it is alleged to contain, and a number of other syntactical, semantic and dialectal problems, all covered by al‑Buʿaymī. As is well known, there has always been disagreement among the Arab grammarians about whether lan was a simple particle, or a phonological variant of lā(laʾ > lan), or a compound of lā + ʾan, and it is remarkable that all Western treatments, with one exception, assume the last to be correct (Lipiński, 366, §39.17, can be taken as a recent representative: he offers no alternative possibilities). The exception is Aartun (1976, 188), who groups it with similar compound elements bearing emphatic suffixes, such as lam, lāta, with ‑n in the case of lan, which he compares with ʾiḏan < ʾiḏā + n. Veccia Vaglieri (2,99) states that lan “stands for lā yakūnu ʾan ‘it will not be the case that’”, via an underlying *lā ʾan, but she offers no evidence of such a historical or formal relationship.

10Western views on the function of lan are influenced by the Arab grammatical tradition. Most agree that lan is an emphatic negative particle assigning a future meaning (e.g. Reckendorf, de Sacy, Veccia Vaglieri, Blachère/Gaudefroy-Demombynes), and some also include perpetuity in their definition, e.g. Wright (2,300) “very strong negation of the future, not at all, never” (Wright’s italics), or else imply it in their translation, as Fleisch (201 n. 1) lan yadḫula “il n’entrera jamais” (italics added). Bergsträsser (15‑17) discusses the range of “meanings” of lan, by which we should understand the different ways it can be translated into European languages, but it is not necessary to pursue them here.

11However we must draw attention to a detail which is usually overlooked: the perpetual sense lies not in lan itself but in the intensifying adverbials which often accompany it. Blachère/Gaudefroy-Demombynes (404) prudently mention these as a separate item after the negative particles, including of course ʾabadan, and Reckendorf (47) deserves credit for reflecting this important distinction in the translation of his two examples, one which does not contain the word ʾabadan, namely S. 3:92, lan tanālū al‑birra “ihr werdet der Frömmigkeit nicht teilhaftig werden”, and one which does, a poetic quotation, lan tahbiṭī ʾabadan, appropriately rendered as “du wirst nimmermehr herabsteigen” (italics added). The same precision is not apparent in Wright’s translation (2,300) of S. 2:24, in which ʾabadan does not appear, fa‑ʾin lam tafʿalū fa‑lan tafʿalū fa‑ttaqū al‑nāra “and if ye do not do it — and ye will never do it — then dread the fire [of hell]” (again italics added). The confusion is widespread: Wright is here fairly close to Sale’s translation (“nor shall ever be able to do it”, Sale’s own italics), and while Sale often does translate lan as purely emphatic with the help of phrases such as “by no means, in no wise”, in Wright’s case theology has been allowed to triumph over syntax in the verse just quoted, with “never” replacing what could have been expressed by ordinary emphasis, “not at all, in no way”.

12In the following discussion lan will accordingly be treated as composite of three elements, (1) negation of the future with (2) emphasis, and (3) according to some, the implication of perpetuity.

13To clear the decks, let it be noted that Sībawayhi deals only with the first item, future negation, defining lan yafʿala as the negative of sa‑yafʿalu or sawfa yafʿalu (Kitāb, Der. 1,87/Būl. 1,68, in §32 and again Der. 1,409/Būl. 1,460, in §259 and Der. 2,332/Būl. 2,305, in §508). This, of course, is repeated in all subsequent grammars, both pedagogical and theoretical.

14But Sībawayhi shows no interest in the emphatic function of lan, which we know, thanks to al‑Buʿaymī, was plainly stated by al‑Ḫalīl b. Aḥmad in the Kitāb al‑ʿayn (8,350, see al‑Buʿaymī, 15): lan is “more emphatic (ʾawkad) than lā”, says al‑Ḫalīl, such that “when you say lan yukrima‑ka zaydun ‘Zayd will not at all treat you generously’, it is as if someone was hoping very much to be treated generously but you denied it and reinforced (wakkadta) your denial by means of lan” (al‑Azharī, who quotes this verbatim in the Tahḏīb, replaces ʾawkad with ʾawǧab “more assertive than lā”, see al‑Buʿaymī, loc. cit.). Sībawayhi makes no mention of this, even though he was well aware of what his master al‑Ḫalīl thought about lan, and indeed rejected his notion that lan was a compound of lā + ʾan (Kitāb, Der. 1,361f/Būl. 1,407, in §233), but apparently other aspects of lan did not come up in their discussions.

15Later grammarians, of course, adopted al‑Ḫalīl’s interpretation (either directly from his Kitāb al‑ʿayn or indirectly through the quotation from it in the Tahḏīb al‑luġa of al‑Azharī, born ca. 280/893, died before 360/970) as part of the definition of lan, and it is sufficient to cite al‑Zamaḫšarī here, lan li‑taʾkīd mā tuʿṭī‑hi lā min nafy al‑mustaqbal “lan is for emphasising the negative future [sense] it acquires from lā” (Mufaṣṣal, 143, §549), and lan naẓīrat lā fī nafy al‑mustaqbal wa‑lākin ʿalā al‑taʾkīd “lan is analogous to lā in negating the future, but with emphasis” (Unmūḏaǧ, 169, §75).

16It might be thought that this definition would be canonical, but in fact both the futurity and the emphasis of lan were challenged, the former on the grounds that the future meaning can also be negated by lā (and lā is indeed classified among the future negatives by al‑Zamaḫšarī himself in Mufaṣṣal, 148f, §578, see Kouloughli’s note 3 to Unmūḏaǧ §86), the latter on the grounds that emphasis is not inherent in lan but is supplied by the context. Both these disputes are well described by al‑Buʿaymī and need not be followed here, as we are only concerned with al‑Zamaḫšarī’s position, which he states with perfect clarity.

17The third element, perpetual negation, has obvious theological implications, which we return to later. At this point it is necessary to review the history of the idea before and after al‑Zamaḫšarī.

18Although the term “Zamaḫšarian lan” implies that it was a creation of al‑Zamaḫšarī, we now know, thanks to al‑Buʿāymī (18ff), that the idea of perpetual negation was already in circulation at least half a century before his time, not unexpectedly in Muʿtazilite circles. Al‑Buʿaymī cites three 5th/11th century authors who discuss or allude to the topic, and, without documenting it, suggests that their ideas could go back to the 4th/10th century, as some of those authors were in the same circle as Abū ʿAlī al‑Fārisī (d. 377/987) and his pupil Ibn Ǧinnī (d. 392/1002), both renowned for their Muʿtazilite leanings, and doubtless open to all the intellectual opportunities of the Abbasid court under the patronage of al‑Ṣāḥib b. ʿAbbād (d. 385/995) and others.

19It cannot be denied that al‑Buʿaymī’s quotations (four in all) are evidence that the connotations of lan were being discussed in broad terms, but taken together they are not conclusive proof that the concept of perpetual negation had achieved scientific grammatical recognition by their time. Two of the quotations mention only the emphatic function (ʾakkada, qaṭʿ, batāt, nafy ʿāmm), and one refers to the future tense without mentioning emphasis at all. Only the quotation from Ibn al‑Ǧabbān (d. after 416/1025, see Sezgin, GAS 8,143, 228f) comes close to the formulation ascribed to al‑Zamaḫšarī, namely: li‑ʾanna lan tufīd nafy al‑šayʾ fī al‑mustaqbal ʾabadan “because lan denotes the negation of something in the future for ever”. But Ibn al‑Ǧabbān is here talking about a poetic hemistich lan yurāǧiʿa qalb‑ī ḥubb‑ahum ʾabadan “my heart will not return to loving them ever”, which contains the word ʾabadan. It is a big leap from interpreting lan accompanied by ʼabadan as denoting perpetual negation to claiming that lan denotes perpetual negation by itself, which is the peculiar property of the “Zamaḫšarian lan”, and al‑Buʿaymī (21) is right to point out that the frequent collocation of lan and ʾabadan must have played some part in the evolution of the idea of taʾbīd al‑nafy.

20Once the idea was in circulation, it was quickly adopted (or challenged!), already by contemporaries of al‑Zamaḫšarī (details in al‑Buʿaymī, 19ff, where he traces the polemic on both sides). We should note especially the phraseology of al‑Ṭabarsī in a work written in 536/1142, a couple of years before al‑Zamaḫšarī’s death, where he speaks of al‑taʾbīd fī al‑mustaqbal and paraphrases the key Qurʾanic verse lan tarā‑nī (S. 7:143) as lan tarā‑nīʾabadan because, as he says, lan tanfī ʿalā waǧh al‑taʾbīd “lan negates in the mode of perpetuity”.

21As al‑Buʿaymī shows, the first to link the concept of taʾbīd with al‑Zamaḫšarī was probably Ibn Mālik (d. 672/1274). The subsequent propagation of the idea was encouraged by Ibn Hišām (d. 761/1360), who is frequently cited by later grammarians, but whether he derived the notion from Ibn Mālik or independently remains to be established.

22Curiously, although Ibn Mālik does name al‑Zamaḫšarī as the proponent of taʾbīd al‑nafy in some of his works (see al‑Buʿaymī, 23‑25 for references), there is one occasion when he appears not to know the name of the author. In his Tashīl al‑fawāʾid, 229, Ibn Mālik states that lan negates the future with or without a limit “contrary to those who make perpetual [negation] a specific feature of lan” (ḫilāfan li‑man ḫaṣṣa‑hā bi‑l‑taʾbīd). What is strange here is that Ibn Mālik mentions al‑Zamaḫšarī in four other places in the Tashīl, also to disagree with him, so it is a mystery why he cannot name him here. Perhaps the Tashīl predates Ibn Mālik’s acquaintance with the Unmūḏaǧ, where taʾbīd is alleged to have appeared. Perhaps the idea was still not universally known: it is significant that an older contemporary of Ibn Mālik, al‑Astarābāḏī (d. 688/1289 or later), while aware of the issue, can only say that “lan is not for long duration and perpetuity as some have said”, laysa li‑l‑dawām wa‑l‑taʼbīd kamā qāla baʿḍu‑hum (or “as someone has said”, Šarḥ al‑kāfiya, 2,218, quoted by al‑Buʿaymī, 17). A little later, in an unpublished commentary on Ibn al‑Ḥāǧib’s Kāfiya by Badr al‑Dīn Muḥammad b. al‑Naḥwiyya al‑Ḥamawī (659/1261‑718/1318, not in GAL) there is a lengthy refutation of taʾbīd (fols. 101r, 101v), but Ibn al‑Naḥwiyya is unable or unwilling to identify any author more specifically than by qāla qawm. The picture is further complicated by the fact that Ibn al‑Naḥwiyya’s wording is the same as al‑Zamaḫšarī’s definition of lan in Mufaṣṣal, §549 (or Kaššāf, 1,349 on S. 7:143), taʾkīd mā tuʿṭī‑hi lā min nafy al‑mustaqbal, but with taʾbīd instead of taʾkīd.

23At all events it can be reasonably assumed that al‑Zamaḫšarī’s name was associated with perpetual negation by the 7th/13th century, but in fairness to him we should look now at his own works. The three most important (not to mention accessible) are his Mufaṣṣal, the Unmūḏaǧ, and his Qurʾān commentary, the Kaššāf.

24The Mufaṣṣal is easily disposed of: in the sections dealing with lan (§§410, 549, 578, 580) there is no trace of taʾbīd. Only the first and second features of lan are mentioned, negation of the future, and emphasis (taʾkīd). The edition of Broch is based on eleven manuscripts with their marginal annotations, two commentaries, and a recently printed edition, yet at no time does he indicate in his critical apparatus that there was a variant reading taʾbīd for taʾkīd anywhere in the text.

25With access to only three versions of the Unmūḏaǧ and four of the commentary of al‑Ardabīlī it is risky to make generalisations about the presence or absence of the reading taʾbīd. What can be said, however, is that if the Unmūḏaǧ is indeed abstracted from the Mufaṣṣal, we would not expect to find taʾbīd there, and this is confirmed by that fact that all the texts consulted have taʾkīd.

26One of them, however, does have a fascinating interlinear correction. In the Saudi MS (PDF p. 27), over the word taʾkīd is written ṣḥ byd, standing for taṣḥīḥ bīd, indicating that an editor has “corrected” taʾkīd to taʾbīd. Since the original text is clearly not that of the manuscript said by al‑Ardabīlī to read taʾbīd, the emendation must have been inserted by someone who knew of that variant, either from another copy of the Unmūḏaǧ or, more likely, from the grammatical tradition.

27This is as far as we can go on the available evidence. We cannot even conclude that there ever was a version of the Unmūḏaǧ with taʾbīd for taʾkīd — we only have Ibn Mālik’s and al‑Ardabīlī’s word that such a manuscript existed (or possibly manuscripts, al‑Ardabīlī’s baʿḍ al‑nusaḫ is ambiguous), and still less can we assume that taʾbīd was al‑Zamaḫšari’s own personal and authorised reading: any copyist could have altered taʾkīd to taʾbīd, as we have seen in the Saudi MS.

28As for the Kaššāf, the general conclusion offered here is vulnerable to the charge than not every page of that work has been checked, nor indeed all of the 105 instances (according to Flügel’s concordance) of the word lan in the Qurʾān. All that can be done is to present some examples of al‑Zamaḫšarī’s treatment of lan taken largely from the verses which attracted the attention of other grammarians and commentators.

29On it’s very first occurrence in the Qurʾān (S. 2:24, Kaššāf, 1,41), fa‑ʾin lam tafʿalū wa‑lan tafʿalū fa‑ttaqū al‑nāra, we are given what must surely be the definitive statement on the difference between lā and lan: lā wa‑lan ʾuḫtān fi nafy al‑mustaqbal ʾillā ʾanna fī lan tawkīdan wa‑tašdīdan “lā and lan are sisters in negating the future, except that lan contains emphasis and intensification”, illustrated by lan ʾuqīma ġadan in answer to someone who refuses to believe the unmarked and neutral lāʾuqīmu ġadan. There is no question here of taʾbīd, the issue for al‑Zamaḫšarī is a binary contrast of futurity (both lā and lan) and emphasis (lan only).

30This distinction is restated in his comment on lā yatamannawna‑hu ʾabadan in S. 62:7, Kaššaf, 2,457, which he compares with lan yatamannaw‑hu ʾabadan in S. 2:95, Kaššaf, 1,67: wa‑lā farq bayn lā wa‑lan fī ʾanna kull wāḥida min‑hā nafy li‑l‑mustaqbal ʾillā ʾanna fī lan taʾkīdan wa‑tašdīdan laysa fī lā “there is no difference between lā and lan insofar as each of them negates the future, except that lan contains an emphasis and intensification which is not in lā”. From this we can safely infer that the only aspects of lan which interested al‑Zamaḫšarī were its future reference and emphasis. Similar pairs occur elsewhere in the Qurʾān, e.g. lan ʾabraḥa in S. 12:80, lāʾabraḥu in S. 18:16, though al‑Zamaḫšarī makes no special remarks on these.

31On S. 5:24, Kaššāf, 1,252, qālū yā Mūsā ʾinnā lan nadḫulahā ʾabadan mā dāmū fī‑hā, al‑Zamaḫšarī points out that ʾabadan here is a taʿlīq li‑l‑nafy al‑muʾakkad bi‑l‑dahr al‑mutaṭāwil “a supplementary comment on the emphatic negation [to indicate] a very long time”. The purpose of the phrase dahr mutaṭāwil can only be to exclude all considerations of negation in perpetuity, at least in the present verse (see further below on S. 19:26).

32The most famous example of the “Zamaḫšarian lan” is in S. 7:143, Kaššaf, 1,348f, where God says to Moses lan tarā‑nī “you shall not see me”. Yet in accounting for the fact that Moses will not see God, al‑Zamaḫšarī makes no mention at all of lan as denoting perpetual negation. Instead he invokes the standard Muʿtazilite doctrine of tawḥīd/tanzīh and asserts that it is logically impossible (muḥāl) and contradictory to God’s attributes (munāfin li‑ṣifāti‑hi) for God to be visible to humans. Significantly al‑Zamaḫšarī again contrasts unmarked lā and the emphatic lan, and also offers a definition of lan in terms very similar to those of the Mufaṣṣal, viz. taʾkīd al‑nafy‑llaḏī tuʿṭī‑hi lā.

33The same Muʿtazilite arguments are used to explain lan yaḫluqūḏubāban in S. 22:73, Kaššaf, 2,67, i.e. that it is logically absurd (muḥāl), and for a fourth time the contrast between lā and lan is invoked, with the taʾkīd feature of lan itself emphasising the logical impossibility of mankind creating anything.

34One of the favourite verses for the opponents of taʾbīd is S. 19:26, Kaššāf, 2,6, when Mary says lan ʾukallima al‑yawma ʾinsiyyan. This is of course a red herring, since al‑Zamaḫšarī would, if he made any comment at all, simply say that this is a very emphatic negative statement, and there is no implied perpetuity, hence no contradiction in the alleged sense of lan “never” and “today”, but he passes over the whole issue in silence.

35These are only the more accessible instances where lan is discussed in the Kaššāf, but they are enough to give a strong impression that al‑Zamaḫšarī never connects lan with taʾbīd. Far from it, his treatment of lan is confined to the two features, negation of the future and emphasis, which he carefully contrasts with those of lā. There are no grammatical properties of lan in the Kaššāf which would justify interpreting it as denoting perpetual negation.

36In the absence of critical editions it cannot be ascertained whether there are any variants in the manuscripts, but in a recent printed edition (Riyadh 1998), based on four manuscripts, there is no variant taʾbīd for taʾkīd in any of the passages cited above. Furthermore, from a searchable PDF version of the Kaššāf it appears that taʾbīd itself is only mentioned in five places, none of them involving lan. Three of them concern expressions intended to imply that something will or will not ever happen, such as the kind of Arab idiom which al‑Zamaḫšarī calls kalimāt al‑taʾbīd, e.g. mā lāḥa kawbabun “as long as a star appears”, comparable with the Qurʾanic mā dāmat al‑samāwātu wa‑l‑ʾarḍu (S. 9:107, Kaššāf, 1,456). The fourth occurs in a discussion of the rejection of the testimony of slanderers (S. 24:4, Kaššāf, 2,84): the Muslims are told lā taqbalū la‑hum šahādatan ʾabadan, paraphrased by Abū Ḥanīfa as ʿalā al‑taʾbīd, and glossed by al‑Zamaḫšarī as meaning “for the length of their lives”. The fifth is associated with the impossible (muḥāl), in this case certain prohibited pre‑Islamic marriage customs (S. 4:22, Kaššāf, 1,199f): al‑Zamaḫšarī equates their impossibility with such phrases as ḥattā yabyaḍḍa al‑qār “until pitch goes white” and ḥattā yaliǧa al‑ǧamalu fī sammi al‑ḫiyāṭi “until the camel goes through the eye of a needle” (S. 7:40), i.e. they can never happen, but lan does not come into the picture anywhere.

37In corroboration we can mention that in his short theological treatise, al‑Minhāǧ fīʾuṣūl al‑dīn, al‑Zamaḫšarī likewise restricts himself to the doctrine of tawḥīd/tanzīh, and makes no appeal to taʾbīd al‑nafy when discussing the Beatific Vision (Eng. text 16f, Ar. 54f). And although he does allude to the use of lan in S. 2:24 lan tafʿalū during his discussion of prophethood (Eng. text 43, Ar. 80), he cites it only as proof that God knows the future, and no taʼbīd is mentioned.

38We can also cite the lemma ʾabad in al‑Zamaḫšārī’s Asās al‑balāġa, where he gives lā ʾafʿalu‑hu ʾabada al‑ʾābād, and variants, as examples of ʾabad in the meaning of “[not] ever”, i.e. “never”. This confirms what we know from the sources quoted above, that in its future sense lan would be synonymous with lā here, plus emphasis, and that perpetuity resides in ʼabad and not in lā or lan.

39If we had to judge only by the works examined above, it would be impossible to conclude that al‑Zamaḫšarī regarded lan as denoting perpetual negation. The question arises, then, why is his name associated with that feature?

40Since the answer is obvious there is no need to discuss it in detail here. As an outspoken Muʿtazilite, al‑Zamaḫšarī drew upon himself the universal hostility of the orthodox for his theologically challenging interpretations in the Kaššāf. Qualitatively the Muʿtazilite feature that would have been most annoying is the inexorable logic with which he and his fellows demonstrated the impossibility of a number of religious assertions, as if it were not enough that they were God’s word, but needed some rational support, for instance, that it is inconceivable (muḥāl) that man himself could create anything (S. 22:73, Kaššāf, 2,67, see above).

41In another place (S. 4:129, lan tastaṭīʿū ʾan taʿdilū bayna al‑nisāʾi, Kaššāf, 1,231) al‑Zamaḫšarī manages to combine this logical impossibility with another favourite Muʿtazilite theme, that of God’s justice, pointing out that it is impossible (muḥāl) to treat one’s wives equally because of our human weakness, hence to impose it on the believers would imply injustice (ẓulm) on God’s part. Al‑Zamaḫšarī, with total consistency, then adds that even the Prophet was incapable of perfect impartiality towards his own wives, an assertion which would not endear him to the orthodox.

42Among the objectionable doctrines of the Muʿtazilites was their denial of the Beatific Vision, ruʾyat Allāh. However, as we have seen, al‑Zamaḫšarī does not explicitly assign eternal impossibility to the negative particle in lan tarā‑nī (S. 7:143, Kaššaf, 1,348f), or indeed to any other instance of lan. Having nothing else to go on, we can only conclude that al‑Zamaḫšarī’s name was attached to that concept because he was the most prominent Muʿtazilite exegete of his day. By whom, we may never find out.

43The suggestion (al‑Buʿaymī 22f) that the Unmūḏaǧ circulated in two forms, an early version with taʾbīd and a later one with taʾkīd, is plausible but still needs to be verified by examining the extant manuscripts. It would not be the first time: the Kitāb of Sībawayhi existed in an “eastern” and “western” recension, with minor differences at the beginning and end, as a means of protecting the copyright of the licensed transmitters and teachers.

44An alternative view (sources in al‑Buʿaymī, 2, 22) is that taʾbīd is not a motivated change but a simple misreading of taʾkīd (the same point is also made by Bergsträsser, 16), and there is even a variant taʾyīd “corroboration, confirmation” in Ibn Mālik’s paraphrase of al‑Zamaḫšarī in one copy of the Tashīl (229 n. 4, see editor’s introduction, 72). It does not help that taʾkīd and tawkīd appear to be used indiscriminately, thus Mufaṣṣal §549 speaks of taʾkīd but uses the verb wakkada in the same sentence, and while lan is said to indicate taʾkīd in Mufaṣṣal (ibid.) and Unmuḏaǧ (§75), it is called tawkīd in Kaššāf (1,41 etc.).

45Al‑Zamaḫšarī has been the victim of another slander, that he changed the opening words of the Kaššaf from al‑ḥamdu li‑Llāhi‑llaḏī ḫalaqa al‑Qurʾān to allaḏī ‎ǧaʿala al‑Qurʾān, allegedly to protect himself by a kind of Muʿtazilite taqiyya, which was subsequently altered to ʾanzala al‑Qurʾān, perhaps not by al‑Zamaḫšarī himself. The anecdote stems from Ibn Ḫallikān (5,170, cf. al‑Buʿaymī, 23),‎ who lived from 608/1211 to 681/1282, about the same time as Ibn Mālik was spreading the story of taʾbīd in the Unmūḏaǧ, and al‑Buʿaymī (12f, 22f) wonders whether the original taʾbīd might have been diluted to taʾkīd for the same reason that ḫalaqa was replaced by ǧaʿala or ʾanzala.

46This does not seem likely. As Lane shows (76‑85), there is not the slightest proof that al‑Zamaḫšarī ever did write ḫalaqa and then change it to ǧaʿala (a Muʿtazilite code word for ḫalaqa) and then to ʾanzala. Lane examined 110 manuscripts of the Kaššāf and found ʾanzala in every one of them, though he did come across a single manuscript where the word ǧaʿala had been inserted over ʾanzala (78f), rather as taʾkīd was “corrected” to taʾbīd in the Unmūḏaǧ, see above.

47Throughout the Kaššāf al‑Zamaḫšarī makes no attempt to conceal or water down his Muʿtazilite views, even though (as Lane observes) he does not take every opportunity to express them. Here we must disagree with Gimaret (350 n. 9), that in matters of exegesis, “le ‎muʿtazilisme de Zamaḫšarī est fort discret, souvent insoupçonnable”, and with Lane (148) that the Kaššāf is not a full‑scale Muʿtazilite work at all, but has “garnered far more criticism for its Muʿtazilite interpretations or influences than its actual contents rightly deserve”. This is a matter of opinion, of course, and all we can say here is that al‑Zamaḫšarī had the reputation of an uncompromisingly assertive Muʿtazilite: Ibn Ḫallikān (ibid.) portrays him as flaunting his Muʿtazilism, mutaẓāhir bi‑hi “making a show of it”.And it suffices to read the violent refutations of Ibn al‑Munayyir printed in the margin of the edition of the Kaššāf used for this paper (cf. Goldziher, 124), to realise that al‑Zamaḫšarī’s heresies were self-evident to his fellow Muslims, and he would have had nothing to gain from such a trivial alteration of his text as Ibn Ḫallikān accused him of making.

48In any case there are good general reasons for scepticism about both accusations, of taʾbīd al‑nafy and the change of ḫalaqa to ʾanzala: they are entirely irrelevant to al‑Zamaḫšarī, in grammar and in theology, and they are inconsistent with his high intellectual standards. We might charitably assume that taʾbīd was attributed to him by some opponent of his position on the Beatific Vision, and originally referred exclusively to lan tarā‑nī in S. 7:143: al‑Zamaḫšarī’s theological arguments against the Beatific Vision could have led this opponent to the false deduction that lan has the grammatical effect of perpetual negation, a misconception which was later extended to lan in other contexts.

49But since we do not know the originator of this misunderstanding (if that is what it was) we can do no more than speculate. The theological conflict may also have been exacerbated by underlying cultural hostilities — it is well known that non‑Arabs were treated with suspicion even when (or especially when) their mastery of Arabic surpassed that of the native speakers (cf. Trumpp’s introduction to his translation of the Mufaṣṣal, 197f).

50To repeat the caveat made at the beginning, the speculations in this paper might be overturned by any of the works which could not be consulted. A small consolation is that these works, particularly the early commentaries on the Unmūḏaǧ by al‑Makkī and ʿAlī b. ʿAbd Allāh b. Aḥmad Zayn al‑ʿArab, are never mentioned in connection with taʾbīd, either in the secondary literature or the grammatical tradition. All our knowledge goes back to indirect sources, Ibn Mālik, al‑Ardabīlī’s Unmūḏaǧ commentary, and Ibn Hišām. Although the e silentio argument is the weakest type, it may be that al‑Ardabīlī is so widely known precisely because he is the only author who claimed that there was a manuscript of the Unmūḏaǧ with the reading taʾbīd for taʾkīd. It may also be relevant that Fischer, in his critical notes on Broch’s edition of the Unmūḏaǧ, has nothing to say on the matter.

51Western knowledge of al‑Ardabīlī may well depend on the coincidence that his commentary was partially edited by de Sacy in 1829, whence it made its way into Howell’s enormous compendium, and from there into the Western grammars. It is not unlikely that de Sacy himself became aware of al‑Ardabīlī’s commentary through d’Herbelot (923), who chose to mention it in his article on al‑Zamaḫšarī, as the manuscript was in the Paris collection, though there is no reference to the taʾbīd issue.

52The link between perpetual negation and al‑Zamaḫšarī survived in spite of its lack of foundation: al‑Kaffawī (d. 1094/1683) makes the myth official in his dictionary of technical terms (4,161f), where he rejects taʾbīd as an unsubstantiated claim (daʿwā bi‑lā dalīl, probably quoting Ibn Hišām, who uses this phrase in the Muġnī on the same topic, 2,221), and he names al‑Zamaḫšarī as the source. Elsewhere in the Kulliyyāt (5,91) al‑Kaffawī quotes al‑Zamaḫšarī’s distinction between the “two sisters” lā and lan in the exact wording of Kaššāf, 1,41 (on S. 2:24), and, again following Ibn Hišām, he dismisses the idea that lan is more emphatic than lā as yet another daʿwā lā dalīl ʿalay‑hā.

53One of the most intriguing cases of the persistence of the idea of perpetual negation (though not associated explicitly with al‑Zamaḫšarī) occurs in the Baḥṯ al‑Maṭālib by the Maronite Germanius b. Farḥāt (d. 1732), who declares (224) that the meaning of lan is nafy al‑istiqbāl muʾabbadan “the negation of the future perpetually”, illustrated by an unattributed Gospel verse, quoted as lan taḥilla ʾan takūna la‑ka mraʾatun and fully parsed in note 510, where lan is termed a “particle of dependence and negation in perpetuity” (nafy li‑l‑taʾbīd). The verse does not correspond exactly to the current Arabic Bible translation, the nearest being Mark 6:18 lā yaḥillu ʾan takūna la‑ka mraʾatu ʾaḫī‑ka “It is not lawful for thee to have thy brother’s wife” in the King James translation, conspicuously lacking the lan (as does a slightly different translation from 1811, mā yaḥillu la‑ka ʾan taʾḫuḏa mraʾata ʾaḫī‑ka).

54The prejudice dies hard. There are only two textual footnotes in the version of the Unmūḏaǧ edited by al‑Manṣūr (1999), one of which concerns the definition of lan as denoting taʾkīd [sic] (32). Al‑Manṣūr’s comment is hāḏā min iʿtizāliyyāt al‑muṣannif, wa‑Llāhu ʾaʿlam, apparently taking taʾkīd to reflect the Muʿtazilite position on lan. It is true that the taʾkīd function of lan was not universally accepted, but there is no reason to connect it with the Muʿtazilites. Can it be that the editor has mentally replaced the orthodox taʾkīd printed in his text by the heterodox taʾbīd attributed to al‑Zamaḫšarī, or at least has understood taʾkīd in that meaning?