This is a great response to someone nitpicking the use of a latin-derived word.
In case that link ever disappears, here it is, reproduced:
Kikjou wrote: Bacteria is plural of bacterium. Please use is correctly. The same goes for media and medium, which is not in this article but is often misused in scientific writing.
In Latin maybe. And the phrase you are nitpicking is actually "from a bacteria". So if you were anything but pedantic, you would exclaim "that requires the ablative of source! In the singular." Following your logic, the article should read "...from a bacterio". But wait, Latin has no indefinite article, so whether it is "a bacterium" or "a bacterio", the noun phrase is redundant since indefiniteness is presupposed in simple noun forms. But "from bacterio" is neither grammatical English nor comprehensible Latin. And the ablative of source usually employs a preposition, so "from a bacteria" should read "ab bacterio" to be exact.
Problem is: This is not FRICKING Latin. This is a word of Latin origin that has entered into English. Therefore our rules apply. Because if you demand a Latin singular, I demand the proper Latin case, pronunciation, etc. We took the plural form for obvious reasons. Because of the physical size of bacteria, the word became a mass noun and functions as both plural and singular. Same reason for taking "data": it is collective. There are rarely "bacterias". And certainly no "datas". One sheep. Two sheep. Ten sheep. One form, all numbers. It is legal in English - accept it.
I can only assume you are one of those people who thinks the plural of "octopus" is "octopi" as well. Except there is no such word as "octopus" in Latin. The word is "polypus", "Octopus" is from the Greek ὀκτάπους, and the plural of that is ὀκτάποδες. And even if people knew "octopodes" was the true plural, they would say it wrong since the epsilon is not silent. Why? Because it is adapted for usage in the new language. It has no obligations to its old morphology and phonology.
By your inanity, if you ever say the word "cherry" for a single unit, I have every right to chastise you. That word never existed in French. "Cheris(e)" is the singular form that was introduced into English. How dare you impose English conventions of depluralization on it! You will say "Shair-eez" for one piece of fruit, you'll do it in a beret and you'll like it. You doctrinaire dope.
I really wish this written exchange was a physical one, where in the end Kikjou "the doctrinaire dope" was punched in throat. So we could all flick the ashes of our cigarettes into his tears of defeat.
"Problem is: This is not FRICKING Latin. This is a word of Latin origin that has entered into English. Therefore our rules apply. Because if you demand a Latin singular, I demand the proper Latin case, pronunciation, etc."
That's not necessarily true. It seems to me that most IE languages have some sort of mixed rules for Latin loanwords. In Czech, for example, the case endings are Czech (not Latin, as you would have it) but the stem is Latin in it behavior (it's mandatory in the official language norm), including (every pupil's favourite) third declension irregularity (miles, -ites; homo, -inis etc.). Therefore, your "everything or nothing" rant seems unjustified. I'd go as far as to say the all transfers of words from language X to language Y make some sort of compromise on all levels of the language, for all values of X and Y.
The reason you don't say from a bacterio is because there is no ablative case (and accusative for that matter) in English for NOUNS. Hence, that would not be used whatsoever. You see, when we take words from Latin or Greek, we do as a matter of fact take the APPLICABLE cases and only the APPLICABLE cases, and incorporate them into English. It is not Latin rules under which we inflect bacterium to bacteria. It is under English rules. There are tons of words which follow this pattern. We do this because it is inconvenient to say bacteriums, hence we borrow the associated inflection from the language from which the word is derived. I repeat, there are tons of words which follow this pattern. Hence, the plural of locus is loci. The plural of persona, is personae, and the plural of bacterium is bacteria.
Furthermore, mass nouns are nouns which cannot be counted, only measured by a certain unit. Bacteria CAN be counted. Moreover, a datum is an individual piece of information, hence data CAN be counted.
Also, just to point out, there is a Latin word octopus, which was conjured up in the period of New Latin.
Also to say there is no indefinite article is an exaggeration, if you really needed something as an indefinite article for whatever reason and however stupid you could just use unus.
Also sheep is a COLLECTIVE noun, not a mass noun.
Re #3: Claiming that "bacteriums" is inconvenient to say is post hoc rationalizing. The "u" vowel sound there isn't special; it's no more difficult to say than "trams" or "slims" or "balms".
There is a very slight pronunciation-based argument for the "us"->"i" rule, but again it's pretty much negated by all the other English singular nouns which end in an "s", and which we pluralize just fine using standard English rules.
There's no way to justify "personae" at all.
This entire set of Latin-based rules was all wanking from Latin-obsessed intellectuals in the late 19th century. There's no guiding principle behind it, no grand justification. It's just "Latin is cultured, so let's make English act more like Latin".
(I get collective vs mass nouns confused as well, so I'm not surprised the poster got confused.)
I think we are missing the point. Bacterium is indeed the singular of bacteria according to current English dictionaries. I can understand the replier's thoughts about importing Latin and Greek names to English; Kikjou was probably too pedantic, but he/she is essentially right and has no responsibility for the actual form of those words. I believe he didn't deserve such a hateful reply.
Re #5: Counterpoint: they did deserve such a hateful reply.