Warning. I am not a scholar on this topic. This is information I have learned or developed myself in the course of being a nerd on the topic since before puberty. I hope my thoughts align with actual scholars but it’s unlikely.
Transliteration is the process of writing a foreign language in a native alphabet. So, for example, writing Georgian or Inuktitut or Korean (using the Hangul normally) in the Latin alphabet. Its purpose is to allow the native reader to make sense of the sound of the foreign words. To be able, possibly, to repeat them vocally, whether or not they are understood. This purpose is important to the process.
However, when I first started transliterating at the tender age of 13 using a stolen book (yes I stole from the library — I was a voracious consumer of books and my allowance was a dime a week and I was in more ways than this ethically compromised) of Greek stuff I was doing the opposite: I was looking for codes, and using the Greek alphabet as a code. Therefore to compose my native language in a foreign alphabet, the opposite of the usual purpose of the process. I did, however, invent many rules that would seem to align with more correct use.
Later I would spend hours transcribing Tolkein’s wildly inconsistent use of the Tengwar everywhere it was found in my copies of The Lord of the Rings, The Hobbit, and The Silmarillion. In fact I was much more interested in this bit than the stories themselves. And in examining his usage I found many of the rules I invented for myself echoed as well as the discovery that while inconsistent, his methods were not purposeless. Pedantic nerd that I was, I also compiled errata so I could tell where Tolkein was playing and where he was just screwing up. And where he was playing was revelatory as well.
So when transliterating you have a translational choice that depends on your purpose: how much of the alphabet’s power do you bring to bear on your choices? This is necessary because different alphabets have different powers. The Latin alphabet has a lot of letters that perform multiple roles and that compose (ch, sh, &c.). So in translating, say, სახლი, to Latin, there’s a lovely glottal k sound in there. The word is typically transliterated as “sakhli” using the flexibility of English usage of the Latin alphabet as well as quasi-standards to imply with an “h” that we swallow the “k” a little as we speak it. We leverage the Latin alphabet to do something it doesn’t do but that English does apparently allow us to imagine it should. We similarly use “zh” here and there to soften a “z” even though in our own language (say, “azure”) we don’t need it because we aren’t teaching pronunciation with our own spelling of English words. We’re just echoing convention.
So as a kid I wanted to write English words in the Tengwar, the Elvish alphabet, because what could be cooler? It’s calligraphic rather than runic and so clearly much superior to the stonecutter’s alphabet of the Dwarves. Anyway, the question is, how to use it? Consider the word “that”.
Now as a strict substitution cipher, a code, I’d use the Tengwar t, h, a, and t. Simple! But how inelegant! Let us, though baby nerd Brad, think instead like an elf. Or even better, think like a human in Middle Earth: they aren’t transliterating. They are simply using the Tengwar because they have no alphabet of their own! So would a human invent the clumsy “th” structure if they had only the Tengwar to write with? Surely they would not! They’d use the existing Tengwar for “th” and get:
That is, a “th” glyph, a “t” glyph, and a “a” diacritic. I’d use the power of the Tengwar to get my job done. And as a consequence an elf who knows no English can effectively sound out the word.
So this is a thing that bugs me about so many fantasy alphabets: they are built to work as substitution ciphers and not actual alphabets. That is, they are just new shapes for Latin letters as used by English speakers. This is…
Alphabets for other languages evolved to support them. Languages evolved to support the alphabets. They are intimately connected. So a credible fantasy alphabet can’t just be a substitution cipher. Too naïve. It has to have its own rules that are leveraged to create a useful transcription: one in which the native user of the alphabet could sound out the word.
A human in Middle Earth, therefore, would not spell “laugh” that way because they don’t have the history of Latin alphabet usage. They don’t live here. They would spell it “laf” or possibly “laff”. But in the Tengwar. Not this, which zanily uses the Tengwar “gh” glyph as well as an “a” and a bastardized “u”:
“Laff”. Using the correct symbol to double the consonant even. You leverage the foreign alphabet and the foreignosity of it is what’s important. It’s what makes the transliteration interesting. It’s why we’re here playing around like this.
Substitution ciphers are fun but they are a million miles less interesting than actual transliteration. Accept no substitutes in your wacky sci-fi and fantasy constructed alphabets. Make a real alphabet, built to serve a different (perhaps literally) tongue, and wonder how you need to twist it to make it say English. Research how the Hangul works, the history of the Inuktitut which was invented only recently to support an entirely oral language. See what choices are made beyond the Latin alphabet.
Normally I’d just draw the Tengwar myself but I got lazy and used this Elvish engraving tool. Its output is Unicode (yes there’s a Unicode set for Tengwar) so I screen-capped it as it was rendered in my browser.