The Quran attests the word معيش maʕāyiš 'ways of life, livelihoods' twice, Q7:10 and Q15:20. This is the plural of معيشه maʕīšah 'way of life, livelihood'. While this would be exactly the form that I would predict for Quranic Arabic based on its orthography, its form in the reading traditions is rather surprising.
All canonical reading traditions are in agreement that this word is supposed to be read as maʕāyiš. This, however, is in disagreement with the rather regular phonotactics of the forms of Arabic that form the basis of the reading traditions (close to 'Classical Arabic', but less unified). In the phonotactics of the reading traditions sequences of original *āyi and *āwi always shift to āʔi, e.g. *qāwimun 'standing (active participle) > qāʔimun; *qāyisun 'measuring (act. ptc.)' > qāʔisun.
The only exception to that rule is found in the Stem III verbs of medial weak verbs, which through analogy of the perfect form have reintroduced the sequences *āyi and *āwi into the language.
The regular forms would be:
pf. *qāwama impf. *yuqāʔimu
pf. *bāyaʕa impf. *yubāʔiʕu
But because the alternation between w and ʔ and y and ʔ does not really occur in the same paradigm elsewhere in the language, the w and y were analogically restored on the basis of the perfect, yielding the classical form yuqāwimu and yubāyiʕu.
This irregular behaviour of maʕāyiš seems to also come as a surprise to Ibn Mujāhid (d. 324AH/936CE) the first canonizer of the seven canonical reading traditions. While his book kitāb al-sabʕ fī al-qirāʔāt 'The book of the seven among the readings' normally does not comment on part of the Quran when all readers are in agreement on a certain reading, for maʕāyiš he in fact takes the time to comment upon this:
قوله (معيش) كلهم قرأ (مَعَايِشَ) بغير همز. وروى خرجة عن نافع (مَعَآَيِشَ) ممدودة مهموزة. قالا أبو بكر: وهو غلط.
"In the word "معيش" all of them (the 7 canonical readers) read maʕāyiša withou the glottal stop. And Xāriǧa on the authority of Nāfiʕ read it as maʕāāʔiša with an overlong vowel(1) and a glottal stop; But Abū Bakr said he was mistaken."
The very fact that Ibn Mujāhid felt the need to comment on this, despite the fact that all the seven canonical readers are in agreement with each other to read it as such, suggest that he -like me- found this rather surprising and explicitly shows that the expected reading by Classical Arabic grammar is wrong.
What is interesting however, is that this unusual behavior of this word has also become part of Classical Arabic. Sibawayh, the first Arabic grammarian has the following to say about this plural:
ولم يهمزوا مقاول ومعايش لأنهما ليستا بالاسم على الفعل فتعتلا عليه وإنما هو جمع مقالة ومعيشة وأصلهما التحريك فجمعتهما على الأصل كأنك جمعت معيشة ومقولة ولم تجعله بمنزلة ما اعتل على فعله ولكنه أجري مجرى مفاعل.
"And do not apply a glottal stop to maqāwil and maʕāyiš because both of them are not nouns based on a verb, so they are pronounced with the glide, as these are the plurals of maqālah and maʕīšah and a vocalic pattern has been applied to their root; And you pluralize their root as if they were **maʕyišah and **maqwalah and you do not take it as a weak verb, but you pronounce it as a mafāʕil plural."
Here Sibawayh appears to argue that because maʕāyiš is a plural of a noun, it is the reason why the plural does not receive a glottal stop. This however is clearly not in line with the evidence from the rest of the language. Many CaCīCah nouns simply have CaCāʔiC plurals, for example:
madīnah pl. madāʔin 'city'
ḥadīqah pl. ḥadāʔiq 'garden'
kabīrah pl. kabāʔir 'major sin'
qabīlah pl. qabāʔil 'tribe'
Sibawayh's explanation is therefore clearly a post hoc explanation that aims to explain why maʕīšah behaves differently from all other words with this shape.
Where did this come from? I currently have no idea. maʕāyiš looks like a loanword from a dialect that had lost the glottal stop which was incorporated into a dialect that did have a glottal stop.
Another question that comes to mind is what came first, maʕāyiš in Classical Arabic grammar or maʕāyiš in the reading traditions? Both explanations seem possible:
First, maʕāyiš could have been an irregularity in canonical Classical Arabic grammar; which was then later incorporated into the reading traditions. This is certainly possible as Sibawayh's grammar describing this phenomenon predates Ibn Mujāhid's canonization of the reading traditions by well over a century. It could be that after Sibawayh described the normative standard of reading this word as maʕāyiš the reading traditions were changed to account for this. The now "incorrect" reading of maʕāāʔiš may then be a leftover from a period that that was simply the regular reading.
The second option is that Sibawayh felt that maʕāʔiš should be the correct reading, but that his familiarity with the (near) consensus among the not-yet canonized, but certainly well-respected reading traditions to read this word as maʕāyiš caused him to write a special-case rule to account for this irregular plural.
The first of these options, in this case, seems slightly more attractive. As shown in the quoted section from Sibawayh's grammar above, it is not just maʕīšah that has an irregular plural that he comments upon, there is also maqālah 'treatise'. This word does not occur in the Quran, so certainly other 'irregularities' of this type do not have their origin in the Quranic text.
The Reading traditions are full of these idiosyncratic readings which either are not in agreement with Classical Arabic grammar or are in agreement with Classical Arabic grammar, but are irregularities of Classical Arabic.
Forms such as these appear to be the result of a centuries-long negotiation of what Classical Arabic is and isn't, the results of which are not straightforward. The resulting language that ends up becoming Classical Arabic is an amalgamation of all kinds of different forms which probably were never present in a single living dialect of Arabic. Identifying such irregularities (2) is an essential step in better understanding how Classical Arabic came to be, and what linguistic ideal was that the Arab grammarians, as well as those who transmitted the Quranic reading traditions, were aiming towards.
(1) In Quranic recitation, it is regular to pronounce any long vowel before a glottal stop with an overlong vowel. A feature of the language of the Quranic reading traditions that does not seem to be reproduced in Classical Arabic.
(2) I identify several more irregular treatments of the Hamzah in my forthcoming article on the Hamzah in the Quranic text in Orientalia. For example, nabiyy 'prophet', is often read nabīʔ in the reading traditions -- the form that we would expect in a dialect that hasn't lost the Hamzah (like Classical Arabic). Nevertheless, here Classical Arabic agrees with the (irregular) pronunciation of Ḥafṣ as nabiyy.