Romanizing Cthulhu

Torpishev

In order to get back into the groove and get more blogging done, I want to do more posts that don’t have a plan or an end goal. Just an “oh that’s neat” and boom, a post be happening.

This is not, unfortunately, a scenario of late-republic era Rome dealing with gribbly space monsters. It is also not a Old God dating sim: It’s me getting into the weeds about linguistics. Namely, the linguistics of the names of alien gods written by a weird old racist and a bunch of pastiche-writing hacks: I am going to, against all logic and reason, attempt to reverse-engineer the pronunciation of the Great Old Ones by way of how their names are written.

Gods help me.

I’m going to be operating under some core assumptions:

1. All of the names we see as readers are Latin-script romanizations of the prehistoric liturgical language Aklo.
2. All the phonemes of Aklo are pronounceable by human beings.
3. Later scholars and occultists just did an extremely shitty romanization job.
4. The romanized names we see are the work of Miskatonic University scholars working with Aklo texts. They are American, working from around 1840-1940, and have some very strange ideas about what classifies as good orthography. They will be influenced by Greek / Latin / English phonotactics and orthography.
   
5. I am flagrantly disregarding what HPL said about the words being crude approximations of unpronounceable alien names, because if the names were actually unpronounceable and alien no one would be able to identify that they were names or even words to begin with: you’d basically just end up with random onomatopoeia of bodily noises. Which is a great twist thematically - imagine having ants call you an approximation of the sound of your knees cracking. But that would undermine the entire exercise by declaring all of it wholly meaningless, and that’s boring.  Also the undercurrent of “these (not-white) cultists can only bastardize a language they cannot comprehend, what do you mean they have their own language and their own names for things” is best left avoided.
6. Cases where the same letter or cluster can have different realizations will be predictable according to the environment.
   
7. In cases of ambiguity, I will default to whatever I think is the most interesting.
8. Each real-world mythos author is treated as a separate in-universe translator, and they will all have their own personal orthographies: sometimes the same sound will be transcribed two different ways.
9. In cases where I have to decide if a string of letters means X or Y, I will go with the “more canon” option, using the schema found here: HPL -> Weird Tales contemporaries -> Derleth & later writers -> Chaosium
10. Rules and pronunciations can and will be revised and refined as we go along.
11. I will fill out gaps in consonant series as I see fit.
    
12. I am taking this way more seriously than any of these guys did, and can and will ignore any of the preceding rules as I feel like it.
    

I will be using some weird transcription symbols, don’t worry about them.

• <angle brackets> are orthography - the letters used, not the sounds: <pool>
• [square brackets]  are narrow transcription: exactly what sounds are being made: [pʰu:l]
• /slashes/ are broad transcription, which focuses on the most relevant features of the sounds and allows for a certain amount of variability: /pu:l/

And now that I have bored you all senseless with minutiae, let’s get cracking with the big lad himself.

#1: Cthulhu

First not just because he’s the poster boy, but because we actually do get a pronunciation from HPL in a 1934 letter to Duane Rimel.

“The actual sound—as nearly as human organs could imitate it or human letters record it—may be taken as something like Khlûl'-hloo, with the first syllable pronounced gutturally and very thickly. The u is about like that in full; and the first syllable is not unlike klul in sound, since the h represents the guttural thickness. The second syllable is not very well rendered—the l sound being unrepresented.”

And we are off to the races with a bad start because “guttural” does not have a meaningful definition in phonetics. It’s used variously to describe velar, uvular, pharyngeal, and glottal consonants, with a high score on weird racism what with how the velar stops /k/ and /g/ and the glottal fricative /h/ aren’t gutteral, but velar fricatives /x/, /ɣ/, and voiced glottal fricative /ɦ/ are (because they aren’t in English but are featured in languages that Anglophones have commonly been at war with).

The consonant cluster we get is <CTH>, which is clearly meant to ape the Greek “chthonic”. That was pronounced with a /kʰtʰ/ in the Classical period and then shifted to /xθ/ by the time of Byzantium. Neither of those has an l in it anywhere, and his “h represents guttural thickness” isn’t helpful either: it could be /x/, since <KH> is often used for it in Greek-origin words, but he uses <CTH> instead of <KH> in the story. It could be /χ/, and someone on Wikipedia (citing no one but themselves) claims it’s the classic Klingon /q͡χ/. Since we don’t really have any better options at the moment, let’s pivot to the vowels.

The vowels actually give us something to work with. Double-checking my regional phonology, it looks for the time being that Rhode Island Eastern New England English still uses /ʊ/ for “full”. Using <oo> for /u:/ is a time-honored tradition for authors who want to sound like Victorian gentleman writing derogatory things about India, which certainly fits ol’ HPL to a T. That’s an easy rule to establish.

HPL 1: <oo> and <ee> represent /u:/ and /i:/

The second rule we can establish here is even more important: /u/ and /ʊ/ form a tense-lax pair, and in Cthulhu /  Khlûl'-hloo we get /u/ in an open syllable and /ʊ/ in a closed one. This is a handy tool that will help us later.

[Aside: /ʊ/ is a weird phoneme in English; in most environments it unrounded and became /ʌ/ or /ə/ (Thank you FOOT-STRUT split, ya bastard). He did say "about like that in full", which I am more than willing to use as my get out of ʊ free card.]

HPL 2: Tense vowels become lax in closed syllables {a, e, i, o, u} => {æ, ɛ, ɪ, ɔ, ʊ} / _ C {C, #}

Back to consonants: “thickly” of course means jack all: <lh> is typically used to stand in for /ʎ/ or /ɬ/, <hl> is often used for /ɬ/ as well, but the way HPL writes the name in the letter indicates it’s meant as two separate phonemes on a syllable division, so that’s a no go. He says that he left out the prevocalic <l> from “hloo”, which he also left out from <Khlûl>. And then there’s that apostrophe stuck in there - if he was actually writing this in IPA,  that would mark primary stress on <hloo>, but he’s dividing the syllables up with a dash and the apostrophe is attached to the end of  <Khlûl> so I am going to say it’s a glottal stop between syllables.

A dead end, perhaps? If he’s just going to leave out sounds, who is to say how any of this is said? Is this all a doomed enterprise?

Not if we use PATTERN RECOGNITION.

Both of those invisible <l> are stuck between a consonant and a back vowel, which would indicate that the presence of a back vowel regularly changes the sound of the consonant before it. HPL just uses <l> and says it’s “thick”, but I actually lied a bit before:  “Thick” does help us here. It confirms that the <l> is not /[l], but it could be [ɫ] or [ɬ]. The first of those just so happens to be called the “dark L” (it’s the L in “pool” in English) and possesses the feature [velarized]. And there is a language where consonants get velarized before back vowels, spoken by people HPL would have had near the top of his list of scary foreigners worshiping dark gods out in the hills

Irish.

HPL 3: Consonants gain the secondary articulation [velarized] when preceding a back vowel.

(Irish does it for /a/, /o/ and /u/, I am playing conservatively here and going with just the back vowels for now)

The <hloo> would be /hˠu:/ under this schema, but I like a good /ɬ/; /ɬˠ/ is a phoneme so rare the only evidence I can find of it is in reconstructions of Proto-Semitic and possibly Moksha (Wikipedia doesn’t include it, PHOIBLE cites one Russian-language paper from 1993).

Executive Decision 1: <h> = /ɬ/, unless part of a consonant cluster representing another phoneme; /h/ is not part of Aklo's phonology.

HPL 4: <l> is /ɫ/ when following a back vowel.

HPL 5: A glottal stop may be added to break up same-place/similar-manner clusters to differentiate syllables and prevent assimilation.

Which means that, for the time being and pending future revisions, <-ULHU> is the romanization of  /- -ˠʊɫʔ.ɬˠu:/.

I swear this will get interesting eventually. And faster!

#2: Yog-Sothoth

We take a wild 180 from obtuse to relatively easy. The syllables go closed-open-closed and there’s nothing weird about the <y> or the <s>. The  <th> could be either /θ/ or /tʰ/ (since the latter often turned into the former), and we don’t really have any justification for /tʰ/ that outweighs “English writer, he’s using /θ/”.

That is going to be the basis of an extremely convenient rule going forward, but it’s a bit too plain for me. So I’m going to cheat a little and do something weird with the stops.

Many languages around the world distinguish their stops not by voicing (vocal cord vibration) but by aspiration (little puff of hair when you say it). English doesn’t, which is why we say that <p> makes the same sound in “pen” and “spin” (first is aspirated, second isn’t). Mandarin Chinese does, but instead of adding h everywhere (like the Romans did when transcribing Greek) or throwing in apostrophes (curse you Wade-Giles!), Pinyin romanization uses <p> for /pʰ/ and <b> for <p>.

The Nuosu (Northern Yi) language (also in China) (which has a rad script) features a 3-way split between aspirated, plain, and voiced stops, which goes like this: <p, b, bb> => /pʰ, p, b/, and wouldn’t you know it, these mythos names have all three of those inputs. Rounding out the selection I am pencilling in that <pp> or a similar pair follows the old Haida orthography of being an ejective. This is in total violation of the premise that Miskatonic scholars invented this romanization, and I don’t care, that was a flimsy premise anyway.

HPL 6: [consonant] + h = [fricative]; <ph, th, sh, zh, kh, gh> => /f, θ, ʃ, ʒ, x, ɣ/

Executive Decision 2: <p, ph, pp, b, bh, bb> => /pʰ, f, p’, p, v, b/

Which all gives us the final result of /jɔk sˠo.θˠɔθ/

(There's no velarization on /j/ because I cannot find a single example in PHOIBLE)

Executive Descision 3: Only labial and dorsal consonants get velarized by HPL 3

(Friendly reminder that “Thoth” in Greek is spelled Θώθ)

And with one and a half in the can, our current HPL phonology looks like:

Consonants: k, θ, s, ɫ, ɬ, j
Vowels: ɔ, o, ʊ, u, u:

Not much to start with, but we’ll be getting a lot more soon enough.

Something to Consider for Later

Looking through the names specific to HPL, I have noticed a very odd, but potentially very helpful pattern: <e> is almost entirely non-existent outside of Dreamlands locations. So far, I've found it only in the words R'lyeh, Yeb, Yhe, Y'ha-nthlei, Dhole, Nyarlathotep, and Gnoph-keh. Four of those look to be a /ɛ/, one is silent, one as /e/, and  the <ei> could be realized as /e:/, but having only one instance of the latter two at all makes me think that we have a gap in the vowel system. <e> might be a leftover, it might represent something else, or it might just be an extremely marginal phoneme. Something to play around with later.

Bonus Round: Nyarlathotep

This is the exception I mentioned earlier; that -hotep marks this as a clearly Egyptian name, or at least HPL’s poor approximation of one. Well, an hour or so on Wiktionary allowed me to kludge together the following:

• n(j) - negation prefix
• j’rr - (of eyes) to become weak, dimmed, or cloudy
• ‘t - a specific moment or span of time
• ḥtp - to be satisfied or content

Which gives us <nj-j’rr-’t-ḥtp> and a modern Egyptological pronunciation of /ni.ɑ.rɛr.ɑt.ho.tɛp/ (the actual Egyptian would have been radically different, but so long as we don’t call pharaoh's scribes to check the grammar I think we can get away with it.)

 “The moment of undimmed eyes is satisfied.”

Yeah, that checks out.