The Pilcrow, as it is known, originated in the Middle Ages, when scribes did not use space to separate paragraphs from each other. (Space was at a premium, and spacing paragraphs had not occurred to anyone anyway.) Instead, scribes started using a Latin abbreviation to indicate a change of paragraph.
Only it wasn’t P for Paragraph. It was C for chapter (capitulum), with some lines through it.
See:
Brian Collins’ answer is impeccably correct for why pronunciation was not spelled pronounciation after the combination of the Great English Vowel Shift and Trisyllabic laxing (a long vowel three syllables back is shortened, as in insane ~ insanity). But all the answers aren’t really answering why pronunciation is still being pronounced pronunciation.
Let’s look at another instance of trisyllabic laxing: private ~ privacy.
In Britain, privacy has a short vowel, following the same rule as insanity and pronunciation. In the US and Australia, it’s a long vowel.
Is that because the US and Australia are illiterate? Why then, so are the British, when they say pirate ~ piracy, with a long vowel. (I’m told piracy still has a short vowel among laywers.)
No, it’s because trisyllabic laxing has resulted in a lack of morphological transparency. The change in vowel looks wrong: if privacy comes from private, they should sound the same. And they’ve been made to sound the same again, a few centuries later, by analogy.
Well, just as piracy has come to sound more like pirate by analogy, and privacy has come to sound more like private, now it’s pronunciation‘s turn. People are tempted to pronounce it pronounciation, for the same reason. I do. And it’s not currently considered correct, but it may yet.
As I commented in https://www.quora.com/Why-isnt-G…
FWIW, if you look in terms of common stems (someone actually went through Pokorny’s etymological dictionary and did the stem counts: The Distribution of Indo-European Root Morphemes ), Armenian has the most common stems with Greek. Given that Phrygian was next door to Greek and no other languages from close by have survived, that makes sense.
Yes, lexicostatistics is no substitute for the comparative method. It’s not nothing either.
My language features in textbook examples of words that both mean and look the same, in completely unrelated languages, by coincidence. e.g. The Handbook of Historical Linguistics. Greek mati and Malay mata, eye. Google “mati mata malay greek” for more examples.
As a bonus: Greek meli Hawaiian meli—not accidental.