Why Is French Spelled Like That?
why alphabets aren't phonetic
Consider devising a writing system for the below utterance from scratch:
/ʒə pø paz‿y.ti.li.ze yn ɔʁ.to.ɡʁaf stɑ̃.daʁ/
I'll give you some pointers. The ~ above the a in /stɑ̃.daʁ/ means the vowel is nasalized. When saying a nasalized "a", some of the air comes out of your nose, instead of it all coming out of your mouth. (This often sounds like you've sort of said n, but shied away at the last second.) The long ties‿between‿words are liaisons. In this language, when a word begins with a vowel, the previous word bleeds into it. Take for granted that /ʁ/ is r, /ʒ/ is j, and /ɔ/ is ou.
If by "phonetic" we mean "relatively close to IPA, where each phoneme has one grapheme," we might end up with something like this:
Je peu paz'ytilize yn ourtograf stɑ̃dar.
Even here, we've had to make certain concessions to the fact that we're writing in an alphabet, not the IPA. For example, the /y/ in /y.ti.li.ze/ and /yn/ aren't quite the same phonemic quality. But nevertheless, pretty satisfactory. Now, what happens if we make a few changes to the verb's conjugation and noun's number? (Pay special attention to the second and last word.)
/nu pu.vɔ̃ paz‿ytilize lɛ.z‿ɔʁ.to.ɡʁafs stɑ̃.daʁd/
Nu puvõun pas'ytilize lez'ourtografs stɑ̃dard.
The problems with our new spelling are coming into focus. If you didn't already know that peu and puvõun are conjugations of the same verb, it would be extremely difficult to figure that out. Even now, with the data you have, you probably can't divine the infinitive.
Moreover, we've drifted further and further away from a strictly phonetic spelling. We now have three different phonemes all represented by "e" — /ɛ/, /ə/, and /e/. If a phonetic alphabet is one where each grapheme is one sound, with no deviation, even slight deviations from IPA make it non-phonetic.
Now, here's a question. If the standard way of spelling the last word was "stɑ̃dar," and you needed to make it plural, what would you have to do? Simple: Remember that its plural form is -d. Each noun and adjective would have its own seemingly random plural form that you would need to memorize.1
By the way, here's the standard way of spelling these sentences:
Je peux pas utiliser une orthographe standard.
Nous pouvons pas utiliser les orthographes standards.
In the French orthography, standard is spelled with the final -d, even when it's not pronounced. When the noun it's modifying is plural, the writer adds an -s, for reasons traditional pedagogy usually describe as "agreement." What does agreement mean? It means that -d is actually pronounced. The -s is never pronounced, except to make a liaison. In spoken French, -d is the sound that marks that the word is plural. But this would be decidedly inelegant in written French, and so we tell the white lie that the -s makes an adjective plural.
Anglophone learners sometimes see this as the frustrating tendency of French to be non-phonetic. But French spelling is no less phonetic than English spelling. Both orthographies made different compromises in different places, to compensate for the fact that a purely phonetic script would be hard to read and hard to learn.
By retaining an older spelling, the orthography is able to show genealogical relationships between words much more readily than a purely phonetic script ever could. Linguists speak of "deep" and "shallow" orthographies: "shallow" orthographies are more phonetic and less etymological, and vice versa.
Among other benefits, a script which shows these relationships is easier to decipher. I bet you, as an English speaker, could puzzle out the meaning of orthographe standard much more easily than ourtograf stɑ̃dar or ɔʁ.to.ɡʁaf stɑ̃.daʁ, even though the latter spellings are more accurate renderings of what a French person actually says.
To a point, anyway. Even IPA has certain conventions in how it's transcribed, and in the case of English, the standard describes an outmoded style of speech. That's the problem with writing something entirely phonetically — phonetics change constantly, from generation to generation and from class to class. If we were to write in an entirely phonetic script, we would first need to decide whose pronunciations are correct and then decide to never change the way we speak. The problem is apparent.
Let's return for a moment to the verbs — je peux and nous pouvons. In the first year of French as a second language, learners memorize three different conjugation patterns (in addition to some irregular forms like être). The -x appended to peux serves no phonological purpose. In fact, it's a homophone with peu, i.e. a little bit. But the -x doesn't merely differentiate peux from peu: it also tells any reader with at least a bit of French learning under their belt that this is the first-person singular form of an -ir verb. All of these little innovations make a proper orthography easier to read.
To state the obvious, alphabets are more phonetic than other writing systems, like abjads and logograms, but that doesn’t mean they’re easy to sound out. (The most obvious proof of this is the existence of the International Phonetic Alphabet.) All alphabets require special training in their usage, since each made different compromises between phonological accuracy and historical continuity.
Even though English and French have come to pronounce the words orthography and standard differently, we still spell them the same way. This facilitates more than "this meme transcends language" comments. It means that all Francophones can easily read each other's writing, and historical texts stretching back a thousand years. When working on my undergraduate thesis, I discovered — to my amazement — that Middle French was much easier for me to parse than Middle English. This, even though English is my first language, and my French was nothing to write home about. But that's because the English chroniclers wrote like this:
Now turners Edward ageyn to London his cite,
& wille wite certeyn, who schent has his mone.
Of clippers, of roungers, of suilk takes he questis.
Olde vsed tratitoures ilk at oþer hand kestis.
Peter of Langtoft (died c. 1307)
… and the French chroniclers, like this:
La dame qui pas n’avoit trop grand’joie, fors de ce qu’elle se trouvoit de-lez le roi son frère, s’étoit jà voulu agenouiller par trois ou par quatre fois au pied du roi son frère ; mais le roi ne lui souffroit, et la tenoit toujours par la main droite, et lui demandoit moult doucement de son état et de son affaire ; et la dame lui en répondoit très sagement.
Les Chroniques de Jehan Froissart (died 1405)
Froissart's French is legible, if not terribly pretty prose by today's standards. French Wikisource has a very nice e-Book of it in the original Middle French, replete with a 19th century glossary of words that had fallen out of common usage by then.2 In general, reading it reminds me of a joke I saw under a cover of the House of the Rising Sun in Middle French:
Luckily, humanity has gotten a lot better at speaking French.
In short, this discrepancy in legibility is because French has an older orthography, whereas English has a newer one. Perhaps it can also be said that French has changed less than English in the centuries since these texts were written. Though the above Froissart passage is basically legible, this is the modern English translation of Langtoft:
Now Edward turned again to his city, London,
and will be certain who has harmed his money.
Coin clippers and cheats, and of such dishonest people, he sought out,
and those of the traitorous bent, on the other hand, he banished.
Personally, I was entirely unable to decipher this passage. I could glean just enough to know was relevant to my thesis, and so spent a day and a half reading an e-Book about Middle English in one tab with an Early English dictionary in the other tab. (Being an Internet-age medievalist is beautiful.) After getting nowhere, I asked my thesis advisor to "help me finish it," and so she supplied 55% of the words and, um, the order they're in. This to say, a modern English speaker cannot read an English text from eight centuries ago without special linguistic training. If Langtoft had been using standard spellings (and if we'd kept those spellings), I might have been able to muddle through it.
But the truth is, these chronicles were outliers. The majority weren't written in the local language, but Latin, or at least a local-language-inflected Latin. Since Google Translate has a decent Latin option, these were significantly easier for me to skim than Langtoft. Even better for my purposes, many had been translated into modern English or French.
For a Western premodernist considering learning a language in order to access more sources, Latin has an impressive ROI. For millenia — up until the 19th century in the case of the Hapsburg domain3 — important documents were often written in Latin. Learning Old English opens up a pretty small cache of documents. But Latin's corpus is very broad, and its popularity as an L2 means many other learners — and therefore textbooks, teachers, and university programs — exist to guide curious scholars in the exploration of that literature.
Since all scripts require special training, why not require more training and create a standard literature in the process? A deep orthography helps with this, but using a classical language takes the logic even further. Most Arabic books are not written in the local dialect, but in MSA. This expands their theoretical readership not only from one country to many, but from this time period into the next.
Chinese has an incredibly deep orthography, with individual characters only sometimes containing a phonetic component. This it has in common with hieroglyphs and cuneiform — all logographic systems have at least some phonetic component. In the case of Chinese, though, that phonetic component was codified so long ago that it isn't always phonetic anymore. That's particularly true when the character isn't being used to write Chinese, but Vietnamese or Korean or Japanese.
Once an individual learns characters, however, it opens a new mode of trans-lingual communication. There's a famous story of shipwrecked Chinese sailors in Japan, who, finding themselves without a language in common, had to write characters in the sand.4 French orthography, descended directly from Latin in some cases, straddles a line between being a classical language useful for brushtalk, and a modern language whose sounds are always changing.
It's also possible that the lemma form would be the plural. For this counterfactual, I'm assuming that the French grammarians would keep the traditions set by the Latin ones, but like all counterfactuals, you could assume the opposite.
Incidentally, the Wikisource version was redacted by one Jean Alexandre Buchon, whose publications English Wikipedia describe as "not very scholarly, though they have been of great service to history." Yeowch!
See page 86 of Benedict Arnold’s Imagined Communities for a discussion of the switch from Latin to German, and how it impacted the Hapsburgs politically.