Dr. M. Ibrahim Faaris, in his book صفوة الراسخ في علم المنسوخ والناسخ, pages 27-30, defines naskh. It's not clear whether it's his definition, or the definition made by Shu`la, whose book he introduces. I say that because he refers to contemporary scholars such as Dr. Mustafa Zayd.
He starts with the linguistic definition, which he says is any of the following: to remove, change, replace, copy or transfer.
Then he talks about the conventional definition (الاصطلاح). He says it is "The lifting of a legal ruling with a later legal evidence from the Quran or the Sunna
I submit that this is NOT the conventional definition agreed with by the majority. Shu`la's definition explicitly adds the Sunna as a possible abrogator. This reflects his following of Imaam Ibn Hanbal, rahimahullah, whom he admired immensely.
He further explains that "lifting a ruling" is not literal. It means "severing its relationship with the ruled." He says that a former ruling is reality and that cannot be lifted. I think that he was paving a way for justifying the notion that a verse stayed in the Quran even after its ruling has been abrogated.
Then he says that a legal ruling is a "directive from God, be it a demand, a prohibition or an option
I submit that this is NOT the definition of a legal ruling. Scholars have largely agreed that an abrogatable ruling is either a mandate or a prohibition. I think perhaps he was giving himself leeway to approve more abrogation claims.
Then he says that "lifting" means removal and that excludes manners of speech that are not a removal; things like specification: it doesn't lift a ruling, it only limits it to some of its elements.
Then he says "legal evidence" means that it is not a rational evidence. Abrogation does not apply to rulings established by common sense, e.g., a deceased person, a mentally ill person or an unaware person are not required to pray. But he quotes the hadeeth that confirms that, so I don't quite understand this particular evidence, though I agree with him that only legal rulings may be abrogated, if anything can.
Then he says that "from the Quran or the Sunna" means that other sources, such as consensus and analogy, cannot be used to prove abrogation. Dr. Faaris may have been the one who said that part, because he refers to Az-Zurqaani's book مناهل العرفان, volume 2, pages 72-73, and the book الأصول من علم الأصول, by `Uthaymeen, page 35.
Then he acknowledges that the definition of naskh by the Sahaaba, the second and the third generations was even wider than the linguistic one! They used it to mean elaboration too. He quotes Ibn Taymiya, from his famous book الفتاوى, volume 23, page 272 and volume 14, page 10, saying,
المنسوخ يدخل فيه في اصطلاح السلف العام كل ظاهر ترك ظاهره لمعارض راجح كتخصيص العام وتقييد المطلق
Translation: "Mansookh (the amended) includes, in the definition of the Predecessors, any apparent meaning set aside because of preponderant conflictor, such as specification of a generality and limitation of an unlimited."
إن لفظ النسخ مجمل، فالسلف كانوا يستعملونه فيما يظن دلالة الآية عليه من عموم أو إطلاق أو غير ذلك
Translation: "The word naskh is a brief, for the predecessors used it in regards to any implication of a verse, such as generality, absoluteness, etc."
So, what is the problem then? Clearly the word is a heteronym, understood and used as such by the early Muslims. So, why was it changed? And by whose authority?
Shu`ba, or perhaps Dr. Faaris, answers that question by attributing the change to Imaam Ash-Shaafi`i. He quotes him, from his famous book الرسالة, pages 106, 110 and 115-116, saying that a necessary outcome of naskh is leaving compliance with the mansookh (abrogated) and a mandate to comply with the naasikh (abrogator).
Dr. Faaris says that Muhammad Abu-Zahra, in his book أصول الفقه, page 146, confirmed that it was Ash-Shaafi`i who made the distinction and that the earlier scholars did not. Then came scholars who followed Ash-Shaafi`i's definition, such as At-Tabari, An-Nahhaas and Dr. Zayd. He says that the latter two were able to refute much of abrogation claims thanks to that new definition!
Does that mean that scholars have a license to change language as they please?
Isn't it more proper, and more scholarly, to accept what the word actually means and how the people who best knew what it meant have actually used it, and conclude that their narrations are therefore inconclusive evidence that abrogation has ever taken place?
The laudable effort by scholars to reduce the number of abrogation claims was IMHO counterproductive. That's because the larger number of naskh claims, caused by the wider definition of the word naskh, diffuses abrogation. No one claim can be cited for evidence of abrogation in particular, but all may be cited as evidence for naskh. By narrowing the word definition to abrogation only, scholars have made a serious charge against rulings of God, a charge that cannot be substantiated because their basis is an assumption.