Two words that have nothing in common except that they’re near each other alphabetically, they’re so obscure they’re not even in the big Webster’s, and pronouncing them is no easy matter:
Magh: “A member of the (largely Buddhist) people of Arakan, a district on the west coast of Burma (Myanmar), and Chittagong, on the Bay of Bengal.” (OED, 2002 draft entry, which adds in smaller type: “Chittagong was formerly part of the kingdom of Arakan but is now in Bangladesh. The Chittagong Maghs were formerly renowned among Europeans in Calcutta as excellent cooks.”) The OED’s etymology is “< Bengali Mag, Magh, name of the kingdom of Arakan, the kings of Arakan and its people, esp. as coastal pirates < Sanskrit Magha a non-Aryan country.” (You can read about the Bengali attitudes towards the “Magh” here and some history here.) Just looking at the word as Generic Foreign, you would pronounce it /mag/ (with the vowel of ah), and this is indeed what the OED suggests; on the other hand, it’s from a Bengali word pronounced /mog/ and is so spelled in early citations, e.g. 1599 R. FITCH in R. Hakluyt Princ. Navigations II. 257 “The Mogen which be of the Kingdom of Recon and Rame, be stronger then the King of Tippara, so that Chatigan or porto Grande is oftentimes vnder the king of Recon” (where Recon is Rakhaing, the local name of Arakan, Tippara is Tippera or Tripura, a hill district of Bangladesh with its own language, and Chatigan is Chittagong, known in Portuguese at the time as Porto Grande; if anyone can tell me what is meant by Rame, I will be much obliged). Furthermore, the short a is pronounced in Hindustani (Hindi/Urdu), the main local language of the British Raj, as a central vowel (like the vowel in cut), which gives us the form mugg under which we find it in Hobson-Jobson, whose entry includes the following judicious observation:
It is beside the question of its origin or proper application, to say… that the Arakanese disclaim the title, and restrict it to a class held in contempt, viz. the descendants of Arakanese settlers on the frontier of Bengal by Bengali mothers. The proper names of foreign nations in any language do not require the sanction of the nation to whom they are applied, and are often not recognised by the latter. German is not the German name for the Germans, nor Welsh the Welsh name for the Welsh, nor Hindu (originally) a Hindu word, nor China a Chinese word.
I presume the word was pronounced to rhyme with bug by those who actually used it in everyday speech, acquiring the current spelling pronunciation only after the fall of the Raj; this is confirmed not only by the quotes in the Hobson-Jobson entry but by the fact that the word is entered in the first edition of the OED as Mug (“The name given in Bengal to natives of Arakan and Chittagong”).
Majorat: The OED draft entry from 2000 defines it as “In France, Spain, Italy, and some other countries: an entailment of an estate by primogeniture; an estate attached to the right of primogeniture” and gives the following etymology:
[< German Majorat (1775) or its etymon French majorat (1701; earlier majorasque (1679)) < major elder (cf. MAJOR a.) + -at -ATE1, after (with change of suffix) Spanish mayorazgo entailment of possessions upon the heir by primogeniture (c1370). The English word sometimes renders other loans from the Spanish word, e.g. Italian maggiorasco (1602; a1587 in form maiorasco), Russian majorat [i.e., mayorat, майорат—LH] (earlier maiorat, < German or Latin). Cf. post-classical Latin maioratus (16th cent. in this sense, prob. also after Spanish).]
(The etymology in the first edition was much simpler, tracing it back to French and Latin.) The draft entry gives the pronunciation as either /’madʒərət/ or /’maʒora/, U.S. /’mædʒərət/ or /’mɑʒorɑ/, all stressed on the first syllable; the first edition gave only the fully French /maʒora/. Now, the last citation in the draft entry is from that most wonderful of autobiographies, Nabokov’s Speak, Memory: “The eldest was Dmitri, who inherited the Nabokov majorat in the then Tsardom of Poland.” The question is, how would Vladimir Vladimirovich, that searcher-out and cherisher of obscure words, have pronounced it? The sentence in question is not in the earlier Russian version, Drugie berega, but the latter contains the phrase (near the start of the second paragraph of Chapter Three, Section 1) после облавы в майоратском бору [pósle oblávy v mayorátskom ború] ‘after a battue in a pine forest inherited by majorat’ (a description that has vanished in the later work, since in the interim he had discovered that the heraldic bears that had inspired it were in fact lions—”brownish and, perhaps, overshaggy beasts, but not really ursine”), so he knew the Russian version of the word (cited in the new OED etymology, perhaps on account of the newly added Nabokov quote), but surely he would not have been tempted to pronounce the English word mahyoh-RAHT, despite the fact that that is an accurate reflection of the identically spelled German word that may be the source of both the Russian and English ones. No, he would have taken pains to use the “correct” English pronunciation… but which? Since the word has never, apparently, been a natural, mother’s-knee part of anyone’s vocabulary, he would have consulted the only dictionary to contain it, the OED (first edition), and thus have given it the benefit of his fluent French. I just wish there existed an audio version of the book as read by its exquisitely multilingual author.