Word Nerd: The Plural of “Referendum”
2 months ago
Hola nerds! I’m back!
And I’m doing a totally apolitical post with no agenda whatsoever about the plural of “referendum.”*
You may have encountered this (generally spirited but good-natured) exchange in recent months, for obvious reasons. The English-conforming “referendums” feels like the younger, more freewheeling use, while the Latinate “referenda” seems more apposite to older, more formal modes of speech (although those with a legal background are far more likely to use the former than the latter). So which is it, and why?
The answer is, in fact, “referendums,” but the why is the more interesting bit.
Plurals in English
So the first thing to do is talk about plurals in general, which means talking about inflection (as I have previously). So Old English had a bunch of different ways to decline nouns, so that the plural of stān (stone) is stānas, the plural of sċip (ship) is sċipu, ƿudu (wood) is ƿuda, mann (person) is menn, bōc (book) is bēċ and nama (name) is naman, coming to four plural suffixes and two stem vowel-shifts – and that’s just counting the regular forms. These endings depended on a bunch of implicit rules on gender and context that are largely lost on modern English readers.
But by Middle English, this had simplified so that most nouns pluralised in only two ways: the “strong”† -es ending, turning engel (angel) into engeles, and the “weak” -en ending, turning name (name) into namen. By Modern English, the -en ending has all but disappeared, leaving just a handful of words,‡ while the -es ending (or more usually just -s) has more or less overtaken the language. Since then, pretty much all new nouns that have formed or been imported have adopted the same ending. As of now, aside from the aforementioned -en endings and a handful of oddities that change the stem (teeth, geese, mice, etc.) or don’t change at all (like sheep and fish**), English nouns overwhelmingly pluralise with some form of -s ending.
But then in weigh our eighteenth-century grammarian buddies. They felt that words imported from Ancient Greek, Classical Latin or Modern French should absolutely retain their native plurals: cactus should become cacti, oasis should go to oases, gateau becomes gateaux. And most of these usages stuck, becoming part of that great artificial “proper English” you were taught at school.
And in some cases, I can see it. It may be a practical, even necessary acknowledgement of the complexities of pronunciation; speaking as someone with a lisp, oasises is a hissing nightmare, and the more elegant oases is a welcome tonic. In scientific or technical writing, the original Latin may be exactly intended; I have no objection to the use of bacterium and bacteria in a medical context.
But by and large, it’s a rule that exists simply because it exists; we should, we are told, learn obscure foreign inflections just because our forebears have decided it’s to be done.
The Case Against
But I don’t buy it. Frankly, most of the time, it’s obtrusive and inelegant, and a practice that should arguably be put to rest. For a few reasons:
- It’s classist. Like all grammatical prescriptivism, this is a practice meant to signal the quality of your education. Using Latin and Greek grammatical forms shows that you’ve learned Latin and Greek – or at least learned English from someone aware of the rules of Latin and Greek. These rules were laid down at a time when the wealthy middle-class were becoming increasingly educated, and were keen to show it, to earn recognition among their higher-born peers and to distinguish themselves from their poorer neighbours.
- It’s colonialist! Weren’t expecting that, were ya? But tell me, why is it these specific languages are given this privilege? The word orang-utan comes from Malay; obeying Malay grammatical rules should give us the plural orang-orang-utan. And mongoose comes from Marathi; the plural should be (roughly) mongoosay. The word horde is Polish originally, suggesting the plural hordy. If we’re not going to obey those rules, it seems strange to insist on making an exception for Latin and Greek, and there’s no defence that doesn’t boil down to “we consider these cultures to be better than those cultures.”
- It’s unnecessary. Look, we already have -s (and -es, not to mention -oes, -ies and -ves), -en, teeth, geese, mice, sheep and god knows what else to try and remember, just in English; why the funk do we need to add more to this nonsense? I’m gonna take my daughter to as many aquariums as I want, with no minimums or maximums whatsoever – there are no formulas to this decision – and you just understood every single word of that sentence whether it upset your frail grammatical sensibilities or not.
And heck, even if you’re determined to preserve the Precious Latin, a bunch of these presumed plurals are wrong, for several reasons:
- There’s no plural in the original. Virus, in Latin, is an uncountable noun; it means “venom,” and like other liquids, has no number (if you have a glass of water, and I have a glass twice as large, I don’t have “two waters”; I just have more water). In early use, science writers tried to get viri to stick, but if there’s no plural in the original Latin, there’s nothing to import.
- The word has moved around speech. It’s time to accept that agenda is a singular noun in English. Agendum (“which is to be done”) isn’t a noun but a participle, a repurposed verb serving more as an adjective than anything else. Fancypants clerks would head a list of tasks agenda in Latin (i.e. “[those things] which are to be done”) in the same way we’d write “to do” now, and the word entered English as a singular noun, referring to the list rather than to the tasks. As a new word doing a new job, the plural should naturally be agendas. (We should also accept that data is an uncountable noun in English, rather than a plural. Just move on.)
- The word isn’t actually Latin. What’s the plural of octopus? It’s sure as shit not octopi; the word isn’t Latin originally, but the Greek ὀκτώπους, transliterated into Medieval Latin in about the fourteenth century (you could construct the Latin octopes from Latin source words in parallel to the Greek construction, but the Romans seem not to have done). The Greek plural ὀκτώποδες would be rendered octopodes, but honestly let’s just use octopuses.
Essentially, any loanword that doesn’t directly import the sense and usage of the original probably shouldn’t use Latin grammatical rules under any circumstances; they’ve become new words, doing new jobs.
All the arguments above apply – class, postcolonialism, practicality etc. – but this one is also actually wrong. Referendum is an example of a word moving around speech. Like agenda and memorandum, in the original Latin it’s a participle, a verb (in this case, referō, “I send back”) inflected to a sort-of-adjective (a lot of Latin participles don’t map onto English words well) meaning “which is to be sent back” – that is, sent back to the people, to give their view on. As a participle, it modifies something else, rather than serving as a noun in its own right.
It can be inflected for number, of course, but it doesn’t hold that it would apply to multiple votes. Agenda (“[those things] which are to be done”) is a plural participle in Latin, but a singular noun in English, referring to one to-do list with several items on it. By analogy, referenda (“[those matters] which are to be sent back”) would presumably refer not to holding several public votes, but to holding one public vote, in which several questions are put.
And if we’re not planning on making that distinction, we should probably recognise that using the Latin plural adds nothing to understanding, and just inflect it with English rules.
Ta-ra, and remember to register to vote.
*No, seriously, this isn’t about another referendum, it’s about the word.
†Pay no attention to “weak” and “strong.” The terms were used in German grammar to distinguish types of verbs (whether they change the stem in conjugation, or just the suffix), then mapped onto nouns, then spread into other languages, where they’re used with no particular consistency. We should probably just do away with the terms.
‡If you’re interested: oxen, children,⁋ men, women, brethren and oddly swine (which used to be spelled sweyen).
⁋For extra fun, children is actually a sort of “double plural”; the older plural was childer, using a super-rare -er form, but -er seems to have at some point picked up an extraneous -en, possibly by confusion with brethren (although the -er in brother isn't a suffix at all).
**The singular and plural of “sheep” have been identical since the Old English sċēap, but I have no idea what happened with “fish” (fisċ), which used to perfectly happily take the plural “fishes” (fisċas).