Earlier this week I wrote about chess and check. What about checkmate? The Oxford Dictionary of English Etymology says:
Aphetic - OF. eschec mat = Pr[ovençal] escac mat, It[alian] scaccomatto, etc. - Pers. shāh māt the king is helpless; see CHECK1 and MAT2.
(MAT2 is "lustreless, dull", and I think the ODEE is saying that this word is from the same French source as the mate in checkmate.)
But the AHD says
Middle English chekmat, from Old French eschec mat, from Arabic šāh māt, the king is dead : šāh, king (from Persian shāh; see shah) + māt, died (from earlier māta, to die; see mwt in Appendix II).
So, some disagreement, then. In searching for more info, I found M.E. Moghadam's "A Note on the Etymology of the Word Checkmate" in the Journal of the American Oriental Society (no. 58, pp. 662-664, 1938, the first page is here). Moghadam argues that the Persian derivation is the correct one. He begins
Both the Webster and the Oxford dictionaries derived the word checkmate ultimately from the Arabic al-shāh māta, meaning "the king died." There are several objections to this etymology.
Interestingly this is no longer the case for either the OED or Merriam-Webster. Moghadam asserts that
Every single word connected with the game of chess in Arabic is either borrowed from the Perisan and arabicised or translated from the Persian into Arabic.
Also, "the shāh in chess is not killed and does not die."
According to Moghadam, Persian māt means
"left (without a way to escape)," or "at a loss," or "perplexed"; hence "pressed" and "defeated" [...] This use of the word māt in Persian is not confined to the game of chess, but is used on all occasions and usually means "surprised" or "at a loss."
That it has nothing to do with the Ar. māta is further proved by the evidence in the older Persian manuscripts about chess, where the word used for "being checkmated" (māt shodan or shāh-māt shodan) is given as dar-māndan, māndan being the root of the word māt.
The verb māndan "to remain" (cf. Avestan mān- in Barth. Altir. Wört. 1124) when prefixed with the prepositions dar or vā, and often without any prefix, means "to be perplexed," "to be at a loss," or "to be exhausted." Shāh mānd means "the king is at a loss" or "has exhausted his resources." Māt is the abbreviated form of mānd, and such abbreviated forms are not at all unusual.
I know next to nothing about Persian, but it seems that Avestan mān- is from PIE *men- "to stay" (and here). By the way, Christian Bartholomae's Altiranisches Wörterbuch is on Google Books, but page 1124 is not available for preview. mānd is not in the IEED project's Indo-Iranian inherited lexicon.
His third point is
Moreover, if the word māt in Persian is a loan word from Arabic, it must preserve some trace of its original meaning, "died." But the word in Persian is never associated with death and we should therefore look elsewhere for its etymology.
Undoubtedly what happened was this: the Arabs borrowed the game and its terminology from the Persians. The first element in the compound shāh-māt was already familiar to them, and to it they prefixed the def. art. al-; the second element was unfamiliar. The observed, however, that when the shāh was made māt, the game terminated. They naturally concluded that the shāh was dead, and by the familiar methods of popular etymology connected it with their own verb māta. Then through the Arabic the word was introduced into the European languages.
Since Moghadam's article, and possibly because of it, the Persian derivation has become more generally accepted over the Arabic derivation. It's certainly the cooler of the two.