Are these Nothing but Sorcerers? – A linguistic analysis of Q Ṭā-Hā 20:63 using intra-Qurʾānic parallels

: The seemingly ungrammatical wording of Q Ṭā-Hā 20:63 ʾinna hādhāni la-sāḥirāni has been cause for much debate, both in traditional Muslim sources as well as in modern discussion. This paper sets out to reevaluate the grammar of the various reading that are present by comparing them not against the normative grammar as it is established by the medieval grammarians, but rather by comparing its grammar to other, comparable construction in the Qurʾān. By analyzing this Qurʾānic verse within its intra-Qurʾānic parallels it is argued that the minority reading ʾin hadhāni la-sāḥirāni is the original intended reading of the ʿUthmānic text, while the grammatically problematic majority reading ʾinna hādhāni la-sāḥirāni is to be understood as an approximation to popular non-ʿUthmānic readings. Through the comparison with other verses, it is also shown that we may gain deeper understanding into verses of constructions of the type found in Q al-Ṭāriq 86:4 ( wa-ʾin kullu nafsin la-mā ʿalayhā ḥāfiẓun ) and shed light on some of the competing canonical readings in these verses.


Introduction
One of the famously controversial verses in the Qurʾān in terms of grammar is Q Ṭā-Hā 20:63.1 In the majority reading2 ʾinna hādhāni la-sāḥirāni appears to have an incorrect inflection for case. The expected form is ʾinna hādhayni la-sāḥirāni, as ʾinna governs the accusative. The accusative option is in fact the reading of the canonical Baṣran reader  despite its disagreement with the rasm ‫لسحرن(‬ ‫هذن‬ ‫3.)ان‬ The problems with this reading were clearly understood very early on. The Kūfan grammarian al-Farrāʾ (144-207/761-822) already cites a report which traces back to the prophet Muḥammad's wife ʿĀʾishah, who claims that this is a scribal error in the Qurʾān.4 There is yet another reading, ʾin hādhāni la-sāḥirāni, which is the reading of the now dominant Ḥafṣ transmission from ʿĀṣim (as well as that of Ibn Kathīr, who however has a slightly different nominative dual form hādhānni).5 This secondary reading has frequently been understood as containing the negator ʾin followed by the la-which takes on the function of ʾillā "except." This gives the meaning, "these two are nothing but sorcerers," drawing upon similar phraseology in the Qurʾān that does use ʾillā like ʾin hādhā ʾillā ṣihrun mubīnun "this is nothing but manifest sorcery" (Q 5:110, 6:7, 11:6, 34:43, 37:15). This interpretation is clearly awkward, as there is no precedent for the asseverative particle la-to take on the meaning of ʾillā in the qurʾānic corpus -or, to my knowledge, outside of it -and thus such an understanding comes off as special pleading to resolve a grammatical issue. 6 However, yet another understanding of this sentence is found in the grammatical literature which has much better support within the qurʾānic corpus, namely that this ʾin is to be understood as the ʾin al-mukhaffafah -a short form of ʾinna. In this paper I will argue that this interpretation is in fact the correct one, and that despite being a minority reading, ʾin hādhāni la-sāḥirāni is the intended ʿUthmānic reading of this verse.

The Sisters of ʾinna
The morphosyntactic behavior of ʾinna "verily" is well-known. It is placed at the beginning of a phrase and topicalizes or emphasizes the word it governs. The noun that it governs is placed in the accusative. Usually this noun follows ʾinna directly, e. g. ʾinna ḷḷāha ʿalā kulli shayʾin qadīrun "God is powerful over everything" (Q al-Baqarah 2:20 and passim),7 but it may be interrupted by a prepositional phrase especially if the governed word is indefinite, ʾinnā fī l-qalʿati sijnan "In the fortress there is a prison".8 There are two other particles that have very similar behavior, namely lākinna "but, yet" and the subordinator ʾanna "that" which likewise govern the accusative.9 Wa-lākinna kathīran minhum fāsiqūn "But many of them are sinners." (Q al-Māʾidah 5:81) ʾa-lam taʿlam ʾanna 'ḷḷāha ʿalā kulli shayʾin qadīr "Do you not know that God is powerful over everything?" (Q al-Baqarah 2:106) However, these particles are not always followed by nouns. Instead, a verb may follow as well. In such a case, shortened forms of these particles are used, namely lākin and ʾan. ʾan is followed by a subjunctive verb, which ends in -a in its non-suffixed forms. In Arabic grammatical thought the subjunctive ending -a is considered to be naṣb just like the accusative ending in the noun. Thus, its morphosyntactic behavior is not considered to change. In the case of lākin, however, this behaviour is not retained, and lākin may be followed by either a perfect verb or regular imperfect verb in -u (rafʿ).10 Wa-mā ẓalamnāhum wa-lākin kānū ʾanfusahum yaẓlimūn "We did not wrong them; but they wronged themselves"  wa-mā tashāʾūna ʾillā ʾan yashāʾa 'ḷḷāhu rabbu l-ʿālamīn "But you will not wish it except if God, the lord of the universe, wills it"  However, such shortened forms are not strictly relegated to the preverbal position. They occasionally can occur in front of nouns. In such cases, there seems to be no obvious difference in meaning from the use of lākinna. For example: wa-lākini 'l-rasūlu wa-'lladhīna ʾāmanū maʿahū jāhadū bi-ʾamwālihim wa-ʾanfusihim "But the messenger and those who believed with him fought with their wealth and themselves." (Q Barāʾah 9:88) lākini 'l-ẓālimūna 'l-yawma fī ḍalālin mubīn "But the wrongdoers today are in obvious error." (Q Maryam 19:38) wa-laysa ʿalaykum junāḥun fīmā ʾakhṭaʾtum bihī wa-lākin mā taʿammadat qulūbukum "And you have committed no sin if you err therein but for what your hearts intended." (Q al-Aḥzāb 33:5) wa-lākin man sharaḥa bi-l-kufri ṣadran fa-ʿalayhim ghaḍabun mina 'ḷḷāhi wa-lahum ʿadhābun ʿaẓīm "But those who open their hearts to disbelief, the wrath of God will be upon them, and they will have a great punishment."  There are a good number of cases where the readers disagree whether lākin or lākinna is used, e. g.

ʾIn al-mukhaffafah
In light of the behavior of the closely related particles lākin(na) and ʾan(na) one might expect that ʾinna should also have a short form that is (predominantly) used when it is followed by a verb rather than a noun. Indeed, there are many examples of this throughout the Qurʾān.20 All of them are either followed by the verb kāna, with a predicate marked with the particle la-, or with verbs of perception/consideration like wajada "to consider to be" and ẓanna "to think to be."21 wa-ʾin kānat la-kabīratan ʾillā ʿalā 'lladhīna hadā 'ḷḷāh "It was indeed difficult, except for those whom God guided" (Q al-Baqrah 2:143) wa-ʾin wajadnā ʾaktharahum la-fāsiqīn "We found most of them to definitely be sinners" (Q al-Aʿrāf 7:102) wa-ʾin naẓunnuka la-mina 'l-kādhibīn "We think you are certainly among the liars" (Q al-Shuʿarāʾ 26:168) Above, it was shown that the short forms lākin and ʾan were used before verbs but could also on occasion be used before nouns in the nominative. The Arab grammarians also consider this to be an option for ʾin, and they consider ʾin zaydun la-qāʾimun "Zayd is standing" to be semantically equivalent to ʾinna zaydan la-qāʾimun.22 Also within the Qurʾān there is evidence for such uses (outside of Q Ṭā-Hā 20:63); however, in all cases there is disagreement among the readers about the specific reading. All have in common that they start with ʾin followed by a noun phrase followed by what is variously read la-mā or lammā. La-mā can transparently be understood as the asseverative la-followed by a relative pronoun mā "what" which introduces a relative clause. These can be translated to English, with somewhat awkward copular relative clauses, as I have done below to accentuate the structure, but the mā is probably best understood as being semantically empty. In the following overview I will refrain from giving a translation of the lammā reading, which I will discuss afterwards.
is not implausible, but alternatively one could argue that la-mmā is simply some sort of not-entirely-understood emphatic pronunciation of the same particle la-mā. What one cannot argue for is that it is the particle lammā with the meaning of ʾillā "except."39 When one accepts that ʾinna can occur with lammā, just like ʾin, then it becomes difficult to accept that ʾin must be considered a negator when it occurs with lammā in Q 36:32, 43:35 and 86:4. Indeed, the medieval authorities occasionally cite la-(m)mā in its asseverative use as a possible -and to some, clearly preferableunderstanding of lammā even for the other verses.40 The remaining two readings with ʾin followed by kullan are explained as instances where ʾin, despite being the short form, continues to govern the accusative case.41 These strike me as less preferable readings which were likely imposed rather artificially onto the text to strengthen the similarity to the ʾin kullun … la(m)mā verses discussed above. While there are several cases of ʾin followed by the nominative, within the Qurʾānic corpus there are no other examples where a shortened ʾin continues to govern the accusative. Another proposed view is that kullan takes the accusative because it functions as the object of the following verb la-yuwaffiyannahum "he will compensate them". This seems odd, as kullan is essentially functioning as the head of an asyndetic relative clause, and therefore its case should follow its function in the main clause, not that of the relative clause. Indeed, already Al-Farrāʾ says it is an option he does not like (huwa wajhun lā ʾashtahīhi) because one does not say **ʾin zaydan la-ʾaḍrabu "it is Zayd whom I will surely hit" or **mā zaydan ʾillā ʾaḍrabu "it is none other than Zayd whom I will hit", so it is a mistake with both ʾillā and la-(fa-hādhā khaṭaʾun fī ʾillā wa-fī 'l-lāmi).42 Therefore, to sum up: ʾinna may indeed occur in its short form ʾin just as lākinna and ʾanna do. This form ʾin is clearly attested in front of verbs in the Qurʾān, but also in several rare cases in front of nouns. The construction in which it occurs in front of nouns consistently combines with kull 'all, each' and an asseverative relative clause la-m(m)ā. In one case this exact same construction occurs which uses the long form ʾinna followed by the accusative kullan, confirming that ʾin should not be understood as the negator ʾin, but instead as the shortened form of the topicalizer ʾinna.

Lammā in the meaning of ʾillā in oaths
The reason why the grammarians and exegetes seem to have been tempted to interpret lammā as meaning ʾillā seems to stem from the use of lammā in oaths, where it can be used interchangeably with ʾillā, e. g., nashadtuka bi-llāhi lammā faʿalta "I swear to you by God, you must do it" may also be expressed as nashadtuka bi-'llāhi ʾillā faʿalta "I swear to you by God, you must do it." Such constructions are explicitly invoked, for example, by al-Zamakhsharī (d. 538/1144) in order to defend the understanding of lammā in the meaning of ʾillā in ʾin … ʾillā constructions.43 However this compares two clearly different constructions, and the ʾillā in such constructions performs a different functional role than the ʾillā in ʾin … ʾillā construction.
As Fischer points out, sentences in oaths introduce the wish with ʾin "if" if it is a negative oath and ʾillā for a positive oath.44 An oath like nashadtuka 'ḷḷāha ʾin rimta hādhā 'l-makāna ʾabadan "I swear to you by God that you shall never leave this place!", should literally be understood as "I swear to you by God that if you ever leave this place (the consequences will be dire)." By extension ʾillā in such oaths does not mean "except," but should be parsed as ʾin-lā "if not," thus bi-ḥayātī ʾil-lā ʾanshadtanī 'l-bayta should be understood as "by my life, if you do not recite the verse for me (the consequences will be dire)!" Lammā may be used in such constructions for positive oaths in the place of ʾillā. If anything, lammā here should therefore mean "if not" and not "except" as it ostensibly would in ʾin … lammā constructions when one takes ʾin as the negator. But it seems to me that the lammā in the oath constructions can in fact be understood as an emphatic form of la-or la-mā, thus a sentence like nashadtuka bi-'llāhi lammā faʿalta could be interpreted as "I swear to you by God that you may certainly do it (or that it certainly may be what you will do)!" This construction therefore does not provide compelling evidence that lammā can mean ʾillā in the sense of "except."

These two are certainly sorcerers!
With the use of the ʾin al-mukhaffafah discussed above, it is now possible to return to the famously problematic verse Q 20:63 that we started this discussion with, which can be understood in a new light with the preceding discussion. When one reads this verse as ʾinna hādhāni la-sāḥirāni the lack of accusative inflection of hādhāni is indeed a grammatical anomaly,45 and for this very reason this reading should not be considered the intended reading of the ʿUthmānic text, and neither should ʾinna hādhayni la-sāḥirāni, which is grammatically unremarkable, but disagrees with the rasm.
The remaining reading, however, ʾin hādhān(n)i la-sāḥirāni, only adhered to by Ḥafṣ ʿan ʿĀṣim and Ibn Kathīr among the canonical readers, can be understood as both grammatical and in line with the rasm. This construction should not be understood as an ʾin … ʾillā construction where la-inexplicably takes on the meaning of ʾillā "except,"46 but instead it should be understood as the asseverative ʾinna … laconstruction, using the short form ʾin instead of ʾinna just like in Q 36:32, 42:35 and 86:4. While such an interpretation has fallen out of favor in modern translations, it is well-attested in the medieval works of (grammatical) exegesis. Thus, Q 20:63 should be translated as "these two are certainly sorcerers!"

Is ʾinna hādhāni la-sāḥirāni the lectio difficilior?47
Although hidden in a footnote of an article otherwise almost entirely unrelated to the topic at hand, Nicolai Sinai has once made the argument that the ʾinna hādhāni la-sāḥirāni reading is to be considered the preferred reading, arguing that this reading is grammatically quite difficult to explain, but is nevertheless the majority reading among the canonical readings.48 It would be difficult to explain this widespread preference for an otherwise awkward grammatical construction, where other grammatical options are present, had it not reflected a more original recitation. Sinai therefore points out that it is the lectio difficilior and should be preferred.
45 One might note that hādhān(n)i only ever occurs in the nominative form in the Qurʾān (see Q 20:63, 22:19) and thus might wonder whether the form inflected for case at all. But here one should note that the feminine hātayn(n)i (Q 28:27) does occur in the accusative form, in a syntactic context where this is expected. From this I believe that it is likely that in the Qurʾānic Arabic dialect the dual proximal deictic inflected for case; cf. Van Putten,Quranic Arabic,278;The Damascus Psalm Fragment,[68][69] It strikes me as likely that this awkward interpretation of la-as ʾillā took inspiration from reports of non-ʿUthmānic companion readings that do use ʾillā like ʾin dhāni ʾillā sāḥirāni. Ibn Khālawayh, Mukhtaṣar, 88; Farrāʾ, Maʿānī, 2: 184. 47 I would like to thank Hythem Sidky for his valuable input in helping me think through this section. 48 Sinai, "When Did the Consonantal Skeleton of the Quran Reach Closure? Part II," 519 n41. This argument seems defensible,49 but would initially seem to go directly against the above argument that ʾin hādān(n)i la-sāḥirāni is not only a grammatically correct reading, but also the one intended by the ʿUthmānic text. However, this need not be a contradiction, and it is important to keep in mind that the ʿUthmānic text is not the original composition of the text, but rather a standardization of the Qurʾān. Variant readings associated with this text are the result of a complex interaction between the (frequently) pre-ʿUthmānic oral tradition and an attempt to adhere to this new standard text. In light of this, it seems possible that Sinai's argument in principle holds up when speaking of the original composition of the text, all the while maintaining that the ʿUthmānic text indeed intended ʾin hādhān(n)i la-sāḥirāni.
It is possible to take as the original composition the much more typical ʾinna hādhāyni la-sāḥirāni construction, while acknowledging that there were other viable and probably existing oral variants such as the intended ʿUthmānic ʾin hādhān(n)i la-sāḥirāni and companion readings such as ʾin dhān(n)i ʾillā sāḥirāni "these two are nought but sorcerers",50 and variations thereupon.51 Abū ʿAmr's reading was able to access the non-ʿUthmānic reading ʾinna hādhayni la-sāḥirāni with only a minor intervention into the rasm, and quite a few non-canonical readers did likewise.52 Abū ʿAmr is remarkable among the canonical readers in more readily -although still very rarely -disagreeing with the ʿUthmānic text,53 and therefore him adopting what would have been a popular pre-ʿUthmānic reading that disagreed with the ʿUthmānic text is easily understood.
This brings us to the lectio difficilior, ʾinna hādhāni la-sāḥirāni. While certainly the majority of the readers adhered to this reading and its reading is the most difficult, it does not mean that therefore it must be considered the most probable original reading of the Uthmanic text. This skips over the issue that Qurʾānic readings are the result of an interaction between the oral tradition and the written text. It is clear that the majority of the readers, both canonical and non-canonical, avoided devi-49 Though, only if we accept the grammaticality of ʾinna being followed by the accusative -which I believe is to be doubted. This would not make it the more difficult reading, but the more unlikely reading. 50 Ibn Khālawayh attributes it to Ibn Masʿūd and al-Farrāʾ attributes it to Ubayy ibn Kaʿb. It seems likely that knowledge of these companion readings gave rise to the rather awkward attempts to understand the ʾin hādhān(n)i la-sāḥirāni construction as an ʾin X ʾillā Y construction by the later exegetes. See ibn Khālawayh,Mukhtaṣar,88;2: 184. 51 The lower text of the Ṣanʿāʾ palimpsest, for example shows clear traces of mā dhān(n)i ʾillā sāḥirāni; see Sadeghi and Goudarzi, "Ṣanʿāʾ 1 and the Origins of the Qurʾān," 86. ating from the rasm, and this should be taken into account when evaluating such variants. Assuming that ʾinna hādhayni la-sāḥirāni was the preferred pre-ʿUthmānic reading, the closest reading that could be achieved without violating adherence to the rasm would be ʾinna hādhāni la-sāḥirāni. It seems that readers preferred ʾinna hādhāni la-sāḥirāni -perhaps emboldened by observations that one finds among early grammarians that the Banū al-Ḥārith would not inflect hādhāni for case54 -over the marked construction intended by the ʿUthmānic text (ʾin hādhāni la-sāḥirāni), or the ones that violated the rasm (ʾinna hādhayni la-sāhirāni or ʾin dhāni ʾillā sāḥirāni). The lectio difficilior argument would then only argue for a common popularity of the ʾinna hādhayni la-sāhirāni reading before readers came to be confronted with a rasm that could not accommodate such a reading -leading to the difficult reading.
As a final note, it is worth pointing out that in this model of the competing readings interacting with the ʿUthmānic standard, it is not necessarily obvious that ʾinna hādhayni la-sāḥirāni is to be considered the lectio difficilior (and therefore potior) in relation to other non-ʿUthmānic readings, most notably ones that would have used the ʾin/mā (hā)dhāni ʾillā sāḥirāni construction. These are all perfectly plausible readings with only a very minor difference in meaning and no exegetical value. It does not strike me as possible to recover what the original wording of the composition would have been. I am not sure that such a question even makes sense if the Qurʾān was originally a (semi-)oral composition, which in its early years enjoyed considerable freedom of oral recomposition and reformulation as the famous sabʿat aḥruf hadith seems to suggest.55

Conclusion
It is hoped that this excursus into the syntax of the Qurʾān highlights the importance of using the Qurʾān itself to understand its syntax. The Qurʾān is a corpus on its own, and frequently has constructions that are rare if not non-existent outside of it. By strictly adhering to what can be gleaned from different sections of the Qurʾān, and closely examining the relevant variant readings in such locations, it is possible to come to new insights that are informed not by later grammatical interpretation but by the corpus itself. Nevertheless, it is important to stress the importance of the medieval grammatical-exegetical works while approaching the syntax of the 54 Farrāʾ, Maʿānī al-Qurʾān, 2: 184. 55 For a compelling discussion on the early Qurʾān within a Parry and Lord-style oral composition framework see Dutton, "Orality, Literacy and the 'Seven Aḥruf' Ḥadīth." Qurʾān in this way. While such works often have a maximalist approach to presenting different options, and it is not always possible to get a clear answer from them as to what the preferred understanding should be, they are an essential body of literature that gives a broad overview of different interpretations that are around. Taking a comparative approach to parallel constructions within the Qurʾānic corpus, it becomes possible to decide which interpretation is the most compelling. Doing so, it becomes clear that Q 20:63, which at first seems like a grammatical anomaly can be understood within the attested grammar of the corpus, and turns out to be a rare, but nevertheless understandable formation if read as ʾin hādhāni la-sāḥirāni "these are surely sorcerers!"