Patterns In the Grimoire

Building the Syntax of the World

Good day, good day friends and fellow nerds. I must apologize for my long absence from the lovely Substack. Certain recent personal events had rather interrupted my flow, but I am happy to announce that I am back and ready to get down and nerdy about fantasy languages. 

---

In my recent  articles I have been working on a foundation to help beginning conlangers properly conceptualize grammar.  Of all the parts of a conlang (save perhaps the vocabulary) conlangers spend the most time and exert the most effort on the grammar. Why is this?  Now ,there is, to be sure, much panic and terror in people’s minds around grammar, which various folks have represented to us as being an arcane list of arbitrary, explicit rules. The grammatical knowledge of a child of five, as we have seen from Berko’s experiments, is sufficient to refute this view and render it ridiculous. The children in the study easily recognized xenowords that they had never heard before as belonging to a pattern. Far from being a list of prefabricated rules, grammar in natural languages is the recognition and deployment of linguistic patterns. In a conlang, grammar is the creation and deployment of linguistic patterns. Now, there are no useless elements in any language. Even elements that might not have a transparent meaning--like the words the, or, be, of--nevertheless serve an important function. Every pattern present in any language, thus, makes a particular linguistic function available to its speakers. In other words, the grammar of a language--or the deployment and recognition of linguistic patterns--lets speakers do things with their language. For example, as you recall from my article on the grammar of time, French verbs are interested more in the temporal positioning of an event, whereas Samoan verbs are more interested in an action’s completion, and these languages’ grammars reflect those perspectives. Conlangers spend so much time upon their languages’ grammars because a language is a tool. Without a grammar it can’t do anything aside from name. 

Doing things with grammar

So, knowing that grammar is a tool, and having been exposed a little bit to some of the parts of that tool, the next question is what precisely can this tool do? The hell’s it good for, like, really? Now of course, asking a linguist what grammar can do is a good way to get a dissertation-length lecture, but for our present purposes I will outline three very broad (and very useful) functions of grammar in a language. (Recall that grammar is defined as the recognition and deployment of linguistic patterns)

1. Grammar lets you make a word’s meaning more precise, e.g. the English word to run has “mutated” versions of itself--ran, running, had run--that add an extra layer of temporal specification to the action it refers to. Old English and Classical Arabic have a quantification pattern that lets them specify if there are exactly two of an object.

2. Grammar lets you re-invent new words from old patterns, like what Lewis Carrol did in The Jabberwocky, or what Paul Andersson did in Uncleftish Beholding

3. Grammar lets you distinguish between ambiguous meanings; alternatively, it lets you create ambiguity.. Take the English sentence “I saw the birdwatcher with the binoculars.”This sentence is ambiguous because it’s not immediately clear whether I or the bird watcher has the binoculars. The phrase “with the binoculars,” as it currently stands in the sentence, can either be about the bird-watcher or about my act of seeing. However, applying a simple movement pattern to the phrase to make it “With the binoculars, I saw the bird watcher,” eliminates that ambiguity. Now, this sentence does sound a little unusual, and an English speaker would probably have recourse to another strategy to disambiguate that sentence. However, a language that had an overt distinction between nouns-as-instruments and nouns-as-possessions wouldn’t run into that problem in the first place, since it would be clear from the getgo what the binoculars were being used for.

---

Now before we go any further into theoretical linguistics, I want to draw your attention to exactly how useful these grammatical functions are for world-building. Being able to add precision to words’ meanings permits a fantasy culture to develop their vocabulary in directions peculiar to them. The Great Lakes Merish language, for example has a very complex pronouns system that reflects their very complex social organization. It includes distinctions both between the nature of a relationship and degree of emotional closeness in said relationship. For this conlang, this type of complexity serves a world-building purpose. It enhances immersion (literally, since it’s underwater).

Having a paradigm whereby a fantasy culture can invent new words is also a fantastic way to develop both your conlang and your conworld simultaneously. In my Draconic language, for example, there are a number of terms that are derived from words I had previously come up with. I used a very simple add-something-to-the-root-word pattern, which linguists call agglutination.  The Draconic work klak has the broad reading “to speak.” Thus klak-ich is “a speech”, where the -ich, ending fits the word into a noun-pattern; u-klak is “to tell” or “to speak at”,where the u- prefix changes the direction of the action; and klak-vit is “to ask, where vit means roughly “query.” Deriving new words from preestablished vocabulary gives the language a sense of internal consistency, which is important for a language of conquering dragons, who would likely object to borrowing foreign words into their tongue. 

Finally, the ambiguity-resolving (or ambiguity-creating) mechanism, permits a conlanger to highlight which situations would be most important to their fantasy culture. After all, a language is more likely to disambiguate situations where it deemed precision to be important. As an extreme example in the direction of ambiguity-as-world-building, my in-progress demonic language is based upon a blasphemous ambiguity between polar opposites. My devils use the same words for good and evil; rich and poor; healthy and sick; living and dead. They deploy linguistic patterns to deliberately confuse their interlocutors, as their communication is motivated by malice. In this conlang, it’s the absence of a disambiguating mechanism that contributes to the world-building, as this ambiguity in the demonic vocabulary would explain why deals with devils so often go sour, and in such a deeply ironic fashion. 

“Wouldst thou like to live deliciously?”

As I have said before and will say again, every level of a conlang, from the nitty-gritty grammar to the the down-and-dirty vocabulary is relevant to the world building. We know that grammar is a tool that permits language to do things. Specifically, it makes words’ meanings more precise; it makes new words from old patterns; and it allows for disambiguation. So with those concepts under our belt, the next question is: what kind of a grammar do you want for your conlang? Every part of your conlang is going to work in service to your story, in service to your world, so the question about grammar gets folded into a question about world-building. Before you can ask what type of grammar are you building, you need to answer broadly what kind of world are you building? I divide world-building methodologies into two general types: 

1. Simple and general

2. Complex and precise

(I know, I know that this is an oversimplification, but bear with me if you please)

The simple and general world-building proceeds like a fairy-tale. The complex and precise world-building proceeds like a history. In the fairy tale all the elements of the story (the setting, the characters, the plot) are organized around the journey of the main character. In a history, every element of the narrative has its own reason for being, its own origin story, so to speak. Let me illustrate what I mean. So we all know the story of Rapunzel, the girl trapped in a tower whose long hair let a prince rescue her, a properly weird fairy tale (as they all are when you stop to think about it). But suppose we started to ask questions about the story elements... A question about the setting: Why is Rapunzel in a tower? “Well”, the fabulist replies, “because the prince needed an obstacle to overcome to save her.” But we are not satisfied and we ask a question about the plot: why is her hair so inconveniently long, wouldn’t she have cut it at some point? “Ah you see”, they reply, “I couldn’t very well give the prince no means of overcoming his obstacle, now could I? And besides long hair is symbolic of spiritual power” From the point of view of the fairy tale, thus, the whole world is built up to help tell a story about the main character. However, if we asked the same questions about the same story to a historian, we would get very different answers. Why was Rapunzel in a tower, we ask the historian, and why was her hair so long? “A fascinating query,” she replies “in the land of Fablia, you see,  it was customary for virgins with blonde hair to be given into to the care of a witch, because blonde hair was the closest to white, which in those lands was sacred because of its overarching position on the color spectrum. These girls were deemed both blessed and dangerous, and as such only a wise woman who had forsaken ordinary social bonds could be trusted to properly instruct their divine potential. To take a blade to their hair was naturally considered taboo, and it was not uncommon for such girls to grow locks exceeding ten feet in length.” From the point of view of a history, thus, every story element, in a certain way, is the main character of its own story, because they all have distinct origins, not necessarily in service to the main character. 

The grammar of a conlang is developed under similar considerations: does it want to focus on a single salient point, like a fairy tale; or does it want to elaborate all of its elements like a history? My demonic language is simple and general, a fairy-tale type of conlang. Its main point is that demons are liars, and delight in so being. My Draconic and Merish conlangs are complex and precise. Their goals are to elaborately characterize entire fantasy cultures, down to the merest details. Why is this distinction necessary for conlanging? It is necessary to figure out right from the start how simple or complex you want your conlang to be, so that you won’t get sucked off on some grammatical tangent. There are hundreds of grammatical patterns available across the worlds languages--your conlang is only going to use a few of them to do what it’s made to do. How many are you going to use and which ones--and  why? In order to answer these questions, the beginning conlanger requires some familiarity with the broad linguistic patterns that apply to different parts of speech. I will illustrate these broad patterns in the next section with examples from natural languages and my own conlangs. I consider grammatical patterns as they apply to 4 types of word: naming-words (nouns); relation-words (verbs); modifying words (adjectives and adverbs); function words (particles and adpositionals)

Rapunzel, Rapunzel let down thy hair

---

Patterns in the Grimoire

Naming-words, or nouns

A naming word identifies a potential participant, concrete or abstract, in a situation or event. Meditation, tiger, liberty, fire, are all naming words in that they are things that you can experience, that can eat you, that can inspire you, or can burn you, respectively. In other words they can all participate, as subjects or as objects, in a situation or event. 

These words fit into 4-sub patterns. 

1. Gender

2. Number

3. case/role

4. agreement

Let’s consider these carefully. First off, Gender is a book-keeping pattern, whereby words are separated into different classes, according to how they sound and/or what they mean. (Though, of course, these aren’t the only metrics languages use to classify words) You’ve got languages like French and Spanish with their two genders; and then you’ve got languages like Giguyu with seventeen genders--except, when a language has as many genders as that, we don’t call it gender: we call it noun class. Let me give you an example from some conlangs.

So, for my draconic language, I figured dragons would have some kind of system for dividing up the world, since they would consider themselves superlative beings. I wanted to move away from the binary gender systems of the Romance languages, and so devised a tripartite distinction between animal, vegetable, and mineral. As with masculine and feminine in the Romance languages, these names for the gender patterns only had the loosest connection to the actual physical composition of the objects named, but instead referred to their position in a draconic cosmology, where beings were either Doers, experiencers, or instruments.

1.  Nouns that refer to objects composed of flesh, that have the capacity for experiences, that are unrooted or unattached to a surface, that move of their own power, that are agentive and animate receive the animal gender. Hence, river (xilvk) Kobold (Kovold), elf (alf), wind (kush-kush), light (Rekt), love (luxtul), dreams (fsalsi), all receive the animal gender.  

2. Nouns that refer to objects that have bark or wood , stone  that are constructed by artifice or grow up from the earth, that are experienced rather than experiencing,  that move because they are made to move receive the vegetative gender. Hence, tree (kath), stalagmite (krokukt), chariot (firlug), revolution (gefegrat), receive the vegetative gender. 

3. Nouns that refer to raw materials, to objects that neither grow nor are products of artifice, to alchemical and conceptual primitives like numbers, and transubstantiation receive the mineral gender. The category also accommodates a great number of miscellaneous nouns, and constitutes the largest class Hence, Zero (rood), Iron (dird zelich),teeth (vrit)  and shield (‘ikti)  all receive the mineral gender. 

Gender is not inherently marked on any of the nouns. That is to say, there isn’t an extra bit like the “-a/o” distinction in Portuguese that marks a word as belonging to the animal, vegetable, or mineral gender. Instead, the adjectives are the ones marked for gender. When a word receives some explicit modification to its form to make it fit into a linguistic pattern, we say that that word is marked for that pattern.  So for Draconic, the standard gender-endings are as follows: 

And these endings would combine with adjectives and function words to make the following phrases. The gender markers are in bold. (Note: the ch is pronounced like it is normally English. The x is pronounced like the h in human, or like the German ch. The Apostrophe is a glottal stop, the sound between the vowels in uh-oh. You don’t have to pronounce it to be understood by dragons: you’ll just have a human accent. Oh, and there’s another sound-level-pattern at play that makes the glottal stop disappear in the environment of other consonants--it’s all a system, don’t trouble yourself with it for the time being) Here are the sentences!

1. Gokr Vrit-ukt 

   Claw sharp.veg. Pl.
   Sharp claws

2. ‘Omka     luga-’ich
   Treasure bigmin.s
   Great treasure

3. Vax            vols    klat-ax
   The.an.sing sun  red.an.sing
   The red sun

4. Lax       gyul     svid-ax
   A.an.sing sky  blue.an.sing
   A blue sky

---

Next, what’s number? Number is a quantification pattern that lets speaker of a language specify how many objects or participants are involved in an event. Most languages encode a simple two-way distinction between singular and plural, though this distinction is by no means the only one available. A number of languages--Old English and Classical Arabic Included--have a dual number and Futuna-Aniwa, a minority polynesian language, even has a triple number, encoded in the word taka, which is different from the number-word for three, tolu: SPEC stands for “specifier,” and is (surprise, surprise) a function word that makes other words more specific. TRL stands for “triple.”

a. taka      fare 
SPEC.TRL  house
‘the three houses’ 

b. taka      pepa 
SPEC.TRL  book
’the three books’ 

Dougherty, Janet W. D. 1983. West Futuna-Aniwa: An introduction to a Polynesian outlier language

A certain number of conlangers (myself included) derive great pleasure from playing with number systems. The reason for this is obvious. How a culture counts the objects in its world has direct bearing on how they conceive of the world around them. The simple singular-plural distinction is functionally very useful, and so it would not be surprising to find any language-culture whatsoever using it. But what of languages that have additional number  distinctions? The presence of the dual number would be justified by a fantasy culture for whom pairings of related objects were important. One could imagine inherently dual terms like the words for hands, lips, legs, lungs, which would so be because those objects in nature always appear in pairs. One could also do world-building in the direction of simplicity. A language where there is no pluralizing  pattern, where more-than-one-ness would have to be inferred from the context

---

Up next let’s consider case and role. I am considering these patters together, even though most linguists would insist on a distinction between them, because case and role are both patterns that let languages quantify a participant’s degree and kind of engagement with a situation.  I should offer a word of warning. Case-systems across languages can get disgustingly complicated, so I will give a simple example to go on with. 

My in-progress Ghériala language, designed within the fairy-tale philosophy, intends to capture the mindset of a fantasy culture devoted to the arts of battle. It will be as suited for a roving orcish war-band as for a martial monastery upon a snowy peak. The principle goal of the language, thus, is to provide a precise characterisation of various martial arts “moves.” Of course, I could have gone the dictionary-driven route, and just made new words for every possible technique and stance and takedown, but then I wouldn’t be making an instrument of the language. There would be no systematicity to it, no patterns. No I needed a productive pattern that could, from a few simple base inputs, generate an entire martial vocabulary. Note how right from the getgo grammatical and fantastical thinking blend together. A pattern that generates new words based on prior inputs gives the conlanguage a sense of internal consistency, and for a language that describes a martial art--a martial art is already a very systematic thing in itself--so there’s a deep artistic harmony at all levels of the conlang’s creation. 

So, how did I go about realizing these goals? In the first place, I knew that I would want to build up techniques from base-words. Consulting with my pugilist friends helped me boil down mechanics of violent engagement into three parts:  1) strike  2) grapple 3) block. So striking, grappling, and blocking will the the semantic basis that we use to derive new words for various techniques. The Gheriala words for “strike”, “grapple”, and “block” are as follows: 1) ka, “strike, applied force” 2) ro, “a grab, a grapple” 3) va, “blockade, a shield.” Of course, there is no force applied, without an instrument to do the applying, and here is where the language takes off. You see, what is a “punch,” but “force applied by means of the fist?” and what is a “kick” but “force applied by means of the foot.” If the Gheriala word for “fist” is ko; and the Gheriala word for foot is bli, then all I need to derive the words for “punch” and “kick” is an instrumentation pattern, whereby I could bring ka “applied force” and ko “a fist” together. With that in mind, the next step was simple: I devised a naming-word pattern that differentiated between the roles that such words could occupy in a phrase. 

Gheriala’s case system

The endings are simple, to be sure, but there is no need here to go crazy with the shape and sound of the suffixes. I’m not going for phonological flourishes here but for syntactical efficiency. 

You’ll note that the ending we want (the one that indicates that something was used to do something; the instrumental) is the -a ending. I decided that the instrument should go to the right of the head-word, and then--voila!--the system is up and running , and I can make new words and start specifying a martial system. 

1. Punch
   Ka-    ko-   a
   Force- fist INST

2. Kick
   Ka-      bli- a
   Force- foot- INST

3. Palm punch
   Ka-    mu-   a
   Force palm INST

4. Knife-hand technique
   Ka-     vli-                  a
   Force blade-of-hand- INST

5. Knee-kick
   Ka-     gol-   a

Force- knew INST

6. Leg-grapple
   Ro-   rok- a
   Grab- leg- INST

And i can keep going, ad infinitum. You might be asking why I decided to go with a noun pattern instead of a verb pattern for these terms. The answer has to do with the respective natures of naming-words (nouns) vs relation-words (verbs). A relation-word sets up a situation. A naming-word can do or can have something done to it in said situation. I wanted the martial techniques of Gheriala to be something its speakers could make use of. I wanted the speakers to do a punch, rather than just have them punch. I wanted a more solid terminology so to speak. There’s another more complicated reason, having to do with the X-bar structure of noun-phrases vs verb-phrases, and the difference between complements and adjuncts, but you needn’t worry about that nonsense. 

Keo se vlia fofoliu: she makes circles with her blade

---

Finally we arrive at Agreement. Agreement is a book-keeping pattern whereby words change their shape to better fit a pattern established by words elsewhere in the sentence. Agreement is sort of a meta-pattern that applies to words who’ve already been fitted, so to speak, in prior pattern. Let me give you an example from my own undergraduate research. Nouns, adjectives, and determiners in Portuguese all agree with each other in number and gender. The fem. and masc. stand for feminine and masculine respectively; the s. and pl. Stand for singular and plural.

A                    Batata         nova
The.s.fem      Potato.fem. new fem.
The new potato

Os                  Carros            novos

The.masc.pl.  Car. masc.pl  new.masc.pl
The  new cars

Here the determiner (the word corresponding to “the”) changes shape between “A” and “O” depending on whether the noun belongs to the masculine pattern or the feminine patterns, and the ending of the adjective changes between -a and-o as well, depending on the gender of the noun. This is gender agreement. Number agreement is shown in the second sentence, where the “-s”  attaches to everything-- the noun, the determiner, the adjective, because we’re talking about more than one new car. When designing a conlang, you should ask whether you want your language to feature agreement, and if so, to what extent? Agreement introduces a supplementary level of complexity into your conlang, because individual words are going to have several potential forms, to fit into the established patterns of their language. If your conlang has agreement, you’ll want to ask how many genders or noun-classes are you going to have? What is going to be the world-building pay-off of complexification? Contrariwise, if you decide to eliminate agreement altogether, that decision too can have world-building consequences, as oftentimes, languages lose agreement as a result of interlinguistic contact. Did your language used to have agreement? What in its history caused its loss? 

---

Well, I have been going on for a good long while now, we still have to consider the patterns that apply to relation-words, modifying-words, and function-words, but since each of them is kind their own thing, and complicated enough to write several articles on on their own, I think I’ll take a break for the time being. Let me just say that i am so glad to be back an Substack and writing for my appreciative audience, and I hope that my writings are as helpful to you as I desire them to be. See y’all next time!