How should we pronounce foreign words?
What is different about words like bread, desk, and sword compared to words like burrito, bureau, and karaoke? If you said that the second set of words comes from other languages, you were right.
The process is called “borrowing,” and it happens when one language adds a word (or sometimes a short sound sequence) from another language into its own “lexicon” (that means “word collection/inventory in a language.”) Linguists also call these words “loanwords,” although both terms are pretty funny, because we certainly don’t give them back! All languages borrow words from other languages; this process is part of the larger category of phenomenon called “language contact.” The more bilingual people there are in an area—that is, the more contact between two languages in one spot—the more likely one language is to borrow from the other, although that’s not the only way it happens.
How Do Loanwords Get Pronounced?
Well, this is where things get tricky—and interesting. Let’s look at these three loanwords: tortilla, armadillo, guillotine. We pronounce the two L’s in tortilla like a Y, as they are in Spanish. Yet in armadillo we pronounce them like an L. Now consider guillotine from French: Most English speakers pronounce the two L’s like a Y, as it is in French, too, but not all English speakers!
This spectrum of pronunciation patterns is called “assimilation.” Because English doesn’t usually pronounce L’s like Y’s, the word armadillo, we could say, has been fully assimilated into English. This could be because it has been around long enough for spelling to have influenced how speakers pronounce it.
What Is Assimilation, Exactly?
Well, in order to be used, loanwords must fit the phonological rules of the borrowing language, since a lot of the time, the original words contain sounds that the borrowing language doesn’t have. So, speakers automatically alter the words so they can say them. All native speakers unconsciously follow these rules. For example, in English we don’t cluster consonants V and L, so when we try, we stick a little space between them, out of necessity, and often without perceiving it, like the way we may pronounce the Croatian name “Vlasic pickles” like “vuh-lassik.” But, in Croatian and other languages such as Russian, that combination is common, so the speakers effortlessly blend the two sounds. Some speakers of Asian languages that do not have an L-sound may replace the L in foreign words with an R—a sound they do have, and one that falls into the same phonological category, so it makes sense as a substitute. This is just like the way English speakers often replace a Spanish rolled R with an English R. These substitutions aren’t random or illogical; they're systematic, and governed by rules. These rules—or restrictions—are called “phonotactic constraints.”
What’s really interesting (and more complicated) is that because word borrowing is a process, there are often long stretches in which the loanwords are pronounced differently in different speech communities. Here’s a fun example: Most of us in the U.S. say fillet a lot like French, without pronouncing the T. However, in Australia the word is fully assimilated into Australian English: They say “fill-it”! (Plus, you most likely have heard or will hear English speakers in some U.S. areas say tortilla with that L sound, too!) These variations across speech communities are possible when the words contain sounds that occur in both languages. For example, we say the French loanword cliché fairly close to the original: “klee-shay.” We are able to do this because we have a similar K-sound, L-sound, E-sound, “sh”-sound, and A-sound in English. Yet, we still alter the sounds slightly, and as a result, the word doesn’t sound like a French speaker is saying it.
Here is an interesting side note: While that fillet pronunciation may sound funny to many of us, most people would agree that the collection of dialects in Australia is fairly socially prestigious, meaning English there is not unfairly stigmatized worldwide, and so people are not likely to accuse Australians of ignorant pronunciations (something that unfortunately does get assigned to more stigmatized language varieties). Language and dialect “prestige,” or lack thereof, refers to the bias and stigma that people assign to ways of speaking that are associated with sociological characteristics, such as the financial or educational success of the speakers or the country. Studies about the interaction between society, culture, and language fall in the realm of sociolinguistics.
Even within one single speech community, there are sometimes variations. For example, while many people say the name of the French cathedral like “noder dame,” others say something more like “nochra dahm,” which is a bit closer to the French it came from.
Whether or not loanwords are adapted to the borrowing language or kept more “as is” depends on a variety of factors. One factor is how much people in the borrowing country hear native speakers of the lending language. Another factor is how much the sounds overlap. A third factor is how often the speech community uses the borrowed word. Yet another factor is the prestige of the lending language: Some studies show that speakers try to keep the original pronunciation more if the language is viewed as socially prestigious. (1) Finally, the longer a loanword has been borrowed, the more likely it is to be assimilated, or, at least, to have one agreed-upon pronunciation. For example, alcohol is technically a loanword, but it’s been around so long that we all say it the same way now. (1)