Language: Gnoll

I don't know about y'all, but when I'm naming NPCs and villages and whatnot, I tend to get kind of obsessive about doing it “right". Slapping together a bunch of phonemes until it sounds thematically correct just doesn't work for me. In the past, I've used the strategy of picking a real language, applying some simple sound changes, and using that, so the names can always mean something and have some linguistic consistency. (For instance, the fallen civilization of the reptilian peoples spoke a fricative-heavy version of Hittite.)

This has never worked right for gnolls. I mean, they're hyena-people, so it just feels weird to go, “okay, gnolls speak Portugese" -- gnoll speech should sound like growls and whoops, not like human speech. My general practice has been to just transcribe sounds my dog makes, which gets pretty close, but it's still kind of haphazard and I don't end up with words that mean anything.

So I've decided to fix that. I figure I can throw together a simple grammar & phonology, generate a list of random Gnoll words, and then pull vocabulary from the Swadesh list and stuff that I think is probably relevant to gnoll society. Then I'll be able to generate names for gnoll NPCs, tribes, & whatnot and have them actually mean something and be internally consistent. And I'll put it up here in case anyone else is as weird and obsessive as I am about this sort of thing.

If you just want the final result, here’s the PDF.

The rest of this post is self-indulgent rambling about the reasoning behind the choices I made designing the language.

---

Phonology

First up, for the uninitiated, this here is the IPA consonant chart.

This is a collection of the most common consonant sounds that exist in the world's languages. There are some that are generally left off the chart for some reason (like the clicks), and a bunch of variations on these themes that are expressed by adding diacritical marks or superscript letters.

Now, of course, this is designed for the human vocal apparatus; gnolls are dealing with some different issues. For starters, hyenae don't really have lips, and the images of gnolls available to us don't seem to be any better off in that department. So the sounds that require lip involvement are out.

I'm also trimming that whole dental/alveolar area. Gnolls have big, pointy teeth -- and I don't know about hyena tongues, but my dog has a fairly large tongue that I can only imagine would get in the way of proper articulation... so any consonants that involve the tongue on/near the teeth could easily blend together into a kind of lisping noise. So let's give them /s/ and /z/, with the understanding that it's usually going to be pronounced with a slight lisp; gnolls won't distinguish between /s/ and /θ/. It might also slip into a palatalized /ɕ/, but they'll all be allophones. At this point, I've taken away some of the most common stops, so to make sure they have some, I'll put /k/ and /g/ into the inventory.

That's a pretty small inventory, but that's okay for this language. If we want a language that sounds kind of canine, it should be very much vowel-heavy, so that it sounds like howls. I'm even going to go so far as to say that Gnoll does not allow consonant clusters. However, we do need some growling noises, so that means adding trills. The closest we can get to a growl in IPA is /ʀ/ up there -- you may know it as that French <r> that sounds like you're coughing up a hairball (though, depending on your dialect of French, that noise might actually be /ʁ/). I feel like there'd be differentiation between different growls, so we'll also add /r/ -- that's the Spanish “rolled" <rr>, not the English <r>; the latter is written /ɹ/. I'm also bringing in a third from outside the standard chart -- /ʢ/, the pharyngeal trill that would fit in that blank space to the right of /ʀ/.

Consonant Inventory

s, z, k, g, r, ʀ, ʢ

Now, we're going to want a large vowel inventory. A couple problems with that -- first, based on dog noises, a lot of the vowels we're familiar with just aren't present. Your mileage may vary, but I've never heard a dog articulate /ɪ/ or /æ/, for instance. Second, once again, no lips. That means no rounded vowels, which means our vowel trapezoid is cut in half:

Now, I could say that gnoll snouts provide a longer oral cavity than human mouths and add some extra vowels in on that justification, or I could pick a handful of vowels that I feel are attested in canine vocalizations and differentiate based on length and tone. The second one doesn't involve making up any symbols, so let's do that. I think we can justify the following vowels:

• /i/, as in “meet".
• /e/, as in “play".
• /ɑ/, as in “hot".
• /ɯ/, similar to the vowel in “boot", but don't round your lips while you say it.
• /ɤ/, similar to the vowel in “go" with no lip rounding.

If we want words that sound like a howl, we should have a decent number of different pitches and we should allow very long vowels. So we'll add to our existing “mid" tone a “low" (represented in IPA with a grave accent), a “high” (acute accent), and an “extra-high” (double-acute), for that dog-whistle effect. We will also have long vowels (IPA ː -- like a colon but with triangles) and extra-long vowels (two colon-things). This should give us a pretty sizable inventory.

Vowel Inventory

i, e, ɑ, ɯ, ɤ, ì, è, ɑ̀, ɯ̀, ɤ̀, í, é, ɑ́, ɯ́, ɤ́, i̋, e̋, ɑ̋, ɯ̋, ɤ̋

iː, eː, ɑː, ɯː, ɤː, ìː, èː, ɑ̀ː, ɯ̀ː, ɤ̀ː, íː, éː, ɑ́ː, ɯ́ː, ɤ́ː, i̋ː, e̋ː, ɑ̋ː, ɯ̋ː, ɤ̋ː

iːː, eːː, ɑːː, ɯːː, ɤːː, ìːː, èːː, ɑ̀ːː, ɯ̀ːː, ɤ̀ːː, íːː, éːː, ɑ́ːː, ɯ́ːː, ɤ́ːː, i̋ːː, e̋ːː, ɑ̋ːː, ɯ̋ːː, ɤ̋ːː

To ease my typing burden, let's look at orthography next.

Orthography

Some of these are pretty easy -- /s, z, k, g, i, e/ can just be written as <s, z, k, g, i, e> and I think we can leave it there. Similarly, it's straightforward enough to write /ɑ/ as <a>, /ɯ/ as <u>, and /ɤ/ as <o>, since those are the closest items on my keyboard.

As for the trills -- since /ʀ/ is like French guttural <r> and /r/ is like Spanish <rr>, we can just say /ʀ, r/ are written as <r, rr>. We could take it a step further and say /ʢ/ can be written <rrr>, but there's a whole keyboard full of letters we could use instead; since /ʢ/ is kind of like a rolled /h/, let's just write it <h>.

Vowel length is also easy. Nobody wants to futz around with macrons and double-macrons, so we'll just write /i, iː, iːː/ as <i, ii, iii>, and do the same with the other vowels. I'm pretty sure there's precedent for that.

I'm also going to minimize the number of accents I have to type. Henceforth, an acute accent means “raise the pitch one step" and a grave means “lower it one step". We'll add a rule to the phonotactics that you can only go up or down one step at a time, and say that the least marked option is for a given vowel to retain the pitch of the previous one.

Now, do the gnolls themselves use the Latin alphabet? Probably not. I’d say they had a writing system of their own. In trying to decide what that might be, it occurred to me that if there’s one medium a bunch of pack-hunting carnivores have plenty of, it’s bone. So if I were to design a native Gnoll orthography — which I am not going to do at this time — I would make it look a lot like Oracle Bone Script. But featural instead of ideographic, so that it’s easier to design. You know what, I’ll come back to this.

Grammar

Grammar is somewhat harder to just derive. I’m going to try and make some decisions based on real-life hyenae… let’s see how that goes.

One of the things that is generally known about hyena society is that the spotted hyena is aggressively matriarchal. Because of this knowledge, it’s not uncommon to make gnoll society similarly matriarchal; I do it in my own games. I want to have that reflected in the language, so let’s say that… verbs are conjugated differently based on whether the action is being performed by a female or a male. If the agent of the verb is genderless, then using a feminine conjugation is a sign of respect. Alternately, using masculine conjugations for a non-masculine agent is a form of insult.

As for other grammatical features, let’s borrow some ideas from the languages spoken in the areas of the world where the spotted hyena actually lives.

So mostly Niger-Congo, with some overlaps from the other major African language families. I’ll take large parts of grammar from Niger-Congo, then, and, say, one idea each from Khoisan, Afroasiatic, and Nilo-Saharan.

One of the most prominent aspects of Niger-Congo languages is their rich system of noun classes. It’s kind of similar to grammatical gender in Indo-European languages, except better-executed and more complex, with more of a grounding in reality. For instance, the Bantu languages (a group that includes significant members such as Swahili, Zulu, Xhosa, & Sotho, and which has contributed several loanwords to English) have separate classes for human, animal, plant, animate, inanimate, abstract, locative, &c. Often, plural forms will be listed as a separate class, and multiple disparate noun classes will share the same plural form. For the sake of easier notation, I won’t do that — but I will provide plural forms for each class, and will allow some plural forms to overlap. I think gnolls would like to use the following noun classes:

• Class I: “Female”
• Class II: “Male”
• Class III: “Predator” (given that gnolls are carnivorous hunter-types)
• Class IV: “Prey”
• Class V: “Plants”
• Class VI: “Vermin” (insects and other things that are too small to be worthy prey)
• Class VII: “Dead Things” (hyenae being historically associated with the dead)
• Class VIII: “Supernatural” (spells, demons, gods, &c.)
• Class IX: “Abstractions”
• Class X: “Tools”
• Class XI: “Body Parts”
• Class XII: “Inanimate” (essentially a catch-all for stuff that doesn’t fit elsewhere)

Much like in Swahili and other Niger-Congo languages, we can indicate noun class with a prefix. We’ll also insist that verbs and adjectives agree in class and number with the relevant noun, so Gnoll can be a little more synthetic with a freer word order. Said agreement can be expressed with just a matching prefix, so as not to get too confusing.

Niger-Congo languages also have a variety of suffixes to stick onto verbs with a whole wide range of meanings. We can borrow that system and say that verbs express tense, aspect, and mood through suffixes, and that you can stick as many suffixes onto a verb as you like. The only restriction is that the tense has to come first.

Finally, Niger-Congo languages tend to be SVO — subject, then verb, then object. That’s a fairly familiar sentence structure, since it’s also what we use in English. We’ll make that the default word order in Gnoll, though it’ll be fairly flexible, as noted above.

Afroasiatic languages generally indicate possession with a suffix. We needed a way to indicate possession, so we’ll take that, and maybe add on a few other noun suffixes to keep it company. I feel like in addition to a possessive, pack hunters might like to have an ablative (indicating movement away from), an allative (movement towards), and a vocative (hey, you!).

The Nilo-Saharan family is questionable — there’s some argument over whether it’s a real family or just the languages that didn’t fit into other families — but many of them do share a cool number system. Instead of singular vs. plural, they have singular vs. plural vs. collective — and the collective is the unmarked form. So if you just say “rock”, you mean rocks in general, or stone as a material, etc; if you’re talking about one rock in particular, you have to add the singular declension; if you’re talking about several rocks, you have to add the plural declension. This is super cool, and I’m going to use it.

Khoisan is actually, as far as I can tell, not officially recognized as a single family anymore. Also, sadly, the various language families that used to be considered Khoisan are not doing too hot, and they’re harder to find information on than other African languages. One thing I can tell you about them, though, is that they have some of the richest phoneme inventories in the world — and they’re best known for their use of “click” consonants. I don’t think I want to put clicks into Gnoll — I really can’t imagine a canine mouth articulating those — but we can borrow one of their other rare phonemes. Many of the languages that use ejective consonants as a true phoneme are from this area, and it just so happens that one of the most common ejectives is /k’/. So we’ll add that to our consonant inventory, giving Gnoll two separate <k> sounds. I was initially tempted to just write the ejective as <kk>, but on further consideration, I think it would be much cooler to write it as <q>. So, real quick, here’s an adjustment:

Consonant Inventory

s, z, k, k’, g, r, ʀ, ʢ

Some Quick Phonotactics

• No consonant clusters
• Tones increase or decrease only one step at a time.
• Tones return to “mid” at word boundaries.
• Accepted syllable patterns are:
• V
• VC
• VV (long vowel or diphthong)
• VVC
• VVV (extra-long vowel or triphthong)
• Speaking of diphthongs and triphthongs, you don’t need to worry about boundaries between vowels; gnolls articulate them in such a way that one blends smoothly into another.
• That is, <uu> is never pronounced like in English “continuum”; it’s always a long vowel.
• Similarly, there’s no phonemic difference between an /i/ followed by an /e/ and an /ie/ diphthong.

And here’s the PDF.