I’m not likely to get too deep into the subject of conlanging — constructing languages — in this series. It’s a topic that can literally fill a book on its own (I have several on my shelf), and it rapidly gets eyeball-deep in grammatical concepts that are completely impenetrable to the layperson (is your language going to be nominative-accusative or ergative-absolutive?). If you want to know more about that kind of thing, I recommend you consult a suitably in-depth work by a specialist.
Instead I’m going to talk about my quick-and-dirty method of conlanging. Because the vast majority of the time, all you need is names — and while grammar can play a role in those, you can also get away without worrying about it, and the difference will be barely perceptible.
The secret is phonology.
. . . or maybe orthography, but for our purposes here the two are interlinked. Phonology = the sounds in the language. Orthography = how those sounds are written. And if you’re consistent about those things, and you use them to differentiate one culture from another, you instantly give your world a feeling of solidity and realism.
In the real world, if I mention Elizabeth Tudor, Diane du Poitiers, and Wu Zetian, you stand a very high chance of guessing that those ladies were English, French, and Chinese. Various factors can mess with this — immigration, colonialism, global trade, and so forth — but the underlying principle that different languages produce different kinds of names holds true. That’s why, in the Memoirs of Lady Trent, I have characters with names like Galinke n Rumeme Gbori, Nour bint Ahmad, and Heali’i. Any single name on its own may not register as being distinctive, but as the data points pile up (Nour is joined by Husam ibn Ramiz, Abu Azali, al-Jelidah, and so forth), your reader will get a sense of what a typical Akhian name looks like, even though Akhia is a made-up place. And they’ll know that wherever Gaetano Rossi is from, it’s probably not there.
Conlang websites and instructional books call this a “naming language,” and they give lots of advice on how to create one. But while you can pick which phonemes (sounds) you want your language to have and which you want to exclude and what laws should govern how they’re used — e.g. “this language has R but not L and only allows voiced stops in word-initial positions” — as that example suggests, you rapidly fall down the hole of needing to know some technical details about phonology.
This is why my quick-and-dirty method is to copy the phonology/orthography of real-world languages. I have an embarrassing number of foreign-language dictionaries on my shelf, from the days before Google Translate; I browse the entries until I have a sense of what kinds of patterns are distinctive to that language, then start making up my own words in that style.
There are, of course, pitfalls to this approach. You may, for example, accidentally recreate a real word. This is nigh-impossible to avoid in some languages; Japanese’s consonant-vowel syllabic structure means that a huge number of two-syllable constructions will turn out to mean something — or probably several somethings, given the high frequency of homophones . I recommend checking your invented name against the dictionary before you commit to it, not necessarily to avoid all possible duplications, but to at least make sure you avoid anything embarrassing. Or you might not recreate a specific word, but inadvertently give your name the structure of a past tense verb or something else distinctive. You’re also limited to languages that use the Latin alphabet or else are frequently transliterated; many of the options on Google Translate come up in scripts I can’t read, without even an audio option to let me hear how it sounds. And on a broader worldbuilding level, you risk creating cognitive dissonance in your reader’s mind if you give a name like Connairt to someone whose culture looks more Zulu than Irish.
Depending on the language you look to for inspiration, you also run the risk of alienating your reader with something too “difficult.” Aliette de Bodard has said that for her Aztec-set Obsidian and Blood series, she gave the characters much shorter names than they would have had in reality, because Acatl is much easier for Anglophone readers to manage than Mictlantecuhtli. I’m more willing than the average reader to tackle names unfamiliar to me, but even I choke on something like Kwakwaka’wakw; the indigenous languages of the Pacific Northwest are unrelated to anything I have even a passing familiarity with, giving me no foothold for parsing that name and figuring out how to pronounce it. There’s a temptation to stick with “easy” things, by which I mean the phonology characteristic of Romance and Germanic languages, maybe some Slavic or Greek if you’re feeling adventurous. Bantu languages? Dravidian languages? Mongolian? No way, man. Too weird.
. . . I hope the problems with that “too weird” reaction are reasonably obvious. Naming only with European phonology is a bit like writing only about white characters. The world is much richer and more complex than that, and we don’t do ourselves any services by assuming that readers can’t handle “funny-looking” names. I’m selective in how I construct things, to avoid chucking specific undigestible combos at my readers; bhfaighidh is a real Irish word, but asking someone to cope with that many lenited consonants at once would be a bit much. But that doesn’t mean I haven’t used Irish Gaelic-style names from time to time, and the same goes for the other languages from which I’ve taken my cues.
Which brings us to the final pitfall, and the reason I mentioned orthography above. That verb bhfaighidh? Depending on your dialect, it’s pronounced like one of two English words: “we” or “why.” I am not kidding. The F vanishes, the BH is pronounced like a W, the GH is sort of swallowed and the DH is like a Y — but if you haven’t studied Irish, there’s no way you could guess that. Letters and the sounds they represent don’t stay the same between languages, especially when you consider specific combinations of them. Someone who has never heard French in their life will make something out of that “Poitiers” above, but it ain’t gonna be how a Parisian might say it. I looked at Tibetan for some of the names in Within the Sanctuary of Wings, but (embarrassingly) didn’t realize until far, far too late just how hilariously Tibetan outclasses English in the spelling/pronunciation mismatch department. (Their last spelling reform was in the ninth century.) The language you’re looking at may include sounds (like the Welsh LL) that don’t really exist in English, or make distinctions (like the difference between aspirated and unaspirated P, very important in Mandarin) that aren’t tracked in this language. And even if you know how these things should be said, your reader very well may not, and they’ll make up their own version.
But that last is true of any names you might use, up to and including real names from the real world. Being prepared to have readers mispronounce things is just part of the job.
So while the quick-and-dirty method has its risks, I really do recommend it as a way of deciding how the names of your setting should look and sound. For me, it’s become a core part of my worldbuilding, so that even when the cultures I’m inventing don’t map very closely to real times and places, I still go “okay, these guys are going to have Sumerian-ish names, and hers is going to have Malay phonology, and then for that guy let’s go with Icelandic.” The result is that even in a city with people from many parts of the world, the reader can tell, without me ever telling them, that the differences are there. And that makes it all feel a bit more real.