This is the first document I've written in a couple of years on my constructed language (provisionally named Sêdnoi). I just can't make myself sit down and explain all the rules as they exist in my head, so I've decided instead to give a quick tour of a lot of the main ideas by carrying out a simple task - the morphological derivation of the Sêdnoi word for dragonfly.

I'm hoping that that'll get me on course to write more stuff - either to document stuff that is crystal clear and complete, or to clean up stuff that is obviously a mess. I'm going to attach some notes in red for places that I need to work on next, but feel free to ignore those.

If you find this to be an unreadable mess, thank you for struggling with it and hopefully it will improve over the next few years. I'm hoping that by leaving a lot of mess in, I can avoid getting bogged down in producing something perfect. I hope you think this is a cool project.

• Each clause is a subjective impression, or a mental image.
• Clauses are comprised of four main components:
  • Agent - the thing taking an action in the mental image,
  • Patient - the thing being acted upon in the mental image,
  • Event - the thing that is occurring in the mental image, or what the mental image is a picture of overall, and
  • Oblique - elements of the mental image not participating in actions.
• A word (or phrase) may serve in any of the four roles above - in particular, there is no lexical distinction between nouns and verbs. The means by which words in a clause are assigned into each of these syntactic roles is explained more below.
• Clauses may be joined together to form larger structures, like sentences. Clauses and their combinations form an "algebra of impressions" - attaching impressions together in various ways to create new ones. Coordination and subordination are examples of ways that clauses may be combined.
• Coordination is a link between clauses at the clausal level - linking one full clause with another full clause. Coordination treats clauses atomically and does not take into account the internal structure of each clause.
• Subordination is a link that treats one clause as atomic, but treats the other as having internal structure by attaching specifically to certain words within it.
• Maybe there are also co-subordinate clauses that treat neither clause atomically. Or it treats some aspects of a clause atomically and some as internally structured. I haven't figured that out yet. This will tie in with the question of null anaphora, which is really unresolved in my mind. I really need to nail down a bit better how everything is structured above the clause level.
• Derivational morphology is fossilized syntax - or put another way, there is no derivational morphology fully distinct from normal syntax. In the case of the word for "dragonfly", the derivation of the word involves a full relative clause that is compounded into one word.
  • For this reason subordinate clause morphology is as terse as possible.

• leθíṅ, /​ˌleˈðʲinʲ/
  • Root: θi
  • Syntactic preferences: EOPA
  • Root ranking: probably ~50's
  • Shell: le…ṅ
  • shimmer, opal, iridescent
• bei, /​ˈpei/
  • Root: bei
  • Syntactic preferences: EAPO
  • Root ranking: probably ~20's-30's
  • Shell: ∅
  • strike, blow, sudden, shock
• cŋo, /​ˈ^kŋo/
  • Root: cŋo
  • Syntactic preferences: AEPO
  • Root ranking: probably 50's
  • Shell: ∅
  • skip or bounce, glance off, reflect, glint

• Opal that strikes, shimmering striker
• This meaning is realized using a relative clause - a construction where the word x is "scoped" or qualified by a clause B, where the word x appears within the clause B.
• This might really be too much like a programming language, but I want to take time eventually to nail down this idea. I've called it scoping, instantiation, contextualization, deplatonization. I want to think about how the abstract word "dog" comes to mean "this particular Labrador in front of me" and draw up a set of rules governing how that works, ideally without too much reference to how anybody else thinks it works. Relative clauses will be one step in the process by which an abstract word gets instantiated.
• In this case, x = "leθíṅ", and B = "bei lcŋeθíṅ".

• The clause "bei lcŋeθíṅ" consists of the word "bei", uninflected, and the word "leθíṅ" in an inflected form, "lcŋeθíṅ", /​l̩ˌ^gŋeˈðʲinʲ/.
• The inflection on "leθíṅ" serves to indicate that "bei" occupies the event role, and "leθíṅ" occupies the agent role. It's an example of a dog inflection.

• In the wordlist above, each word root is given a sequence of "preferences" and a (rough) root ranking. These are inherent properties that attach to the root regardless of what word it appears in.
• Consider the clause "bei leθíṅ".
• θi most prefers to occupy the role of event, followed in succession by oblique, patient and finally agent. Bei most prefers to be the event, followed by agent, patient and oblique.
• Since both words first prefer to be the event, there is a "tie", which is resolved in favor of bei as a result of its higher root ranking.
• θi is assigned its second preference, the role of oblique.
• The clause bei leθíṅ would thus read something like "a strike (occurs) among opals".
• Clearly, bei leθíṅ does not have the meaning we desire. We need a way for the word leθíṅ to be read as the agent. This is achieved using a dog inflection.
• A dog inflection allows the preferences of one root to substitute for another. In this case, we substitute the preferences of the root cŋo into the word leθíṅ. This kind of inflection has syntactic content only - the semantics of cŋo don't directly affect the semantics of the inflected word leθíṅ, although using cŋo is an aesthetic choice available to the speaker and a different aesthetic effect could be achieved with the use of a different root.
• In this case (morphology explained more later), the resulting inflected version of leθíṅ is lcŋeθíṅ.

• All relative clauses have a designated word x and a clause B as above.
• Suppose x_R is the root of the word x and its shell is comprised of the parts x_S1 and x_S2, where x_S1 is the fragment preceding the root and x_S2 is the fragment following.
• This is an inexact description of x_S1 and x_S2 as the shell may vary irregularly, but it works for now.
• In this case, x_R = "θi", x_S1 = "lcŋe", x_S2 = "ṅ".
• Suppose B_O is the clause B with the head word (x) omitted.
  • In this case, B_O = "bei".
• The relative clause is created by forming words from x_R, x_S1 and x_S2 and placing them parenthetically around the clause B_O.
• The clause is x_R B_O x_S1x_S2.
• In this case the relative clause is "θi bei lcŋeṅ" - "opal that strikes".
• To treat this simply as a word instead of a full clause, we remove word breaks, adding epenthetic "ê" vowels if these are phonotactically required. The result is *θibeilêcŋeṅ. This is not quite the final form as the stress pattern of this word is now out of balance and will be changed by phonological processes.

• I wrote up a detailed explanation of this in the past, but I've since changed my mind about a few things, so here I go again barbarically fudging it to get the outcome I want.
• Here's the main point, that has never really changed as long as I've been thinking about this language. Every word has a stress pattern that you can think of as "anapaestic".
• So every syllable is one of 3 kinds: a p-syllable taking primary stress, an s-syllable taking secondary stress, or a t-syllable taking tertiary stress (or alternatively, unstressed).
• Within a word, every syllable following a p-syllable is an s-syllable, every syllable following an s-syllable is a t-syllable, and every syllable following a t-syllable is a p-syllable. T-S-P-T-S-P...
• At one time, I thought that every word should have primary stress on either the ultimate or penultimate syllable. Then I stopped really caring about that, and just insist that every word should have primary stress on the root. It's a nice interaction between morphology and phonology.
• We haven't even gotten into the idea of prosodic balancing, but the change mentioned above affects how it would work a bit.
• The idea of prosodic balancing is the following: since morphological processes switch and shuffle syllables a lot, how do we maintain a rigid anapaestic stress pattern? By redistributing vowels around after any morphological process to reinstate it.
• I haven't rewritten the full rules for this but the following is a rough idea of how it would go (I'm twisting it to make the compound word be what I want, and then I'll rewrite the rules to justify that):
• We're starting off with θibeilêcŋeṅ. What needs to change about this word?
• Firstly, note that the root is θi. That has to be a p-syllable, which means the stress pattern of the word is PTSP.
• We now focus on "imbalances", where a p-syllable has a syllable nucleus that is shorter than the nucleus of an adjacent s- or t-syllable, or an s-syllable has a syllable nucleus than that of an adjacent t-syllable. The "length" of a syllable nucleus is ranked with diphthongs as "longest", followed by single vowels, followed by syllabic consonants and the schwa ê.
• Actually, something I forgot to mention. I don't have a phonology section up here, but I envision the "stress" system as being basically pitch accent, with vowel lengths being independent but, through the process of prosodic balancing, still having a certain interaction with stress.
• Write up the phonology section. This part is actually pretty locked in already, if I can resist the temptation to add a new doubly articulated consonant every time I circle back.
• We perform a series of iterated transformations on the vowels of the word until no imbalances remain. Since p-syllables are the most stressed, we resolve P-T imbalances first, then P-S, then S-T.
• Lil vampin'. This could look something like:
• θíbeilêcŋeṅ
• θíbelêcŋeṅ
• θíblecŋeṅ
• Note that at this step we've lost a syllable. The stress pattern has gone from PTSP to PTS. We would continue resolving imbalances according to that new pattern, but we're already done.
• Obviously, a proper write up of imbalance resolution rules. I might leave myself a couple of different options for how each kind of imbalance might get resolved, to give myself a lever to make the etymologies nice. Then once I figure out my own pattern, I can make that the rule.

• θíblecŋeṅ, /​ˈθi.pleˌ^gŋenʲ/
  • Root: θi
  • Shell: …blecŋeṅ
  • I just had an idea, but I don't really know how viable it is. I always wanted the morphology to be primarily infixing, rather than prefixing or suffixing. I guess with this wacky syllable-shuffling paradigm, we might chalk that up as a success already. But here's an observation: if we want an infixing morphology, and for the morphological changes to focus on the root, then ideally we would like words in the lexicon to have their roots smack in the middle. The problem with the derivational paradigm I've explained in this article is that it chronically forces the root to the front of the word, where I don't really like to have it. I have yet to think of a solution to this, and the resolution might be to set down the desire for an infixing morphology altogether. But there's one nice vestige we could possibly hold on to. In a word like θíblecŋeṅ, we could insist that the shell can't all in one part. In this case the first part is empty and the second is blecŋeṅ. Maybe we could create a rule that says, in a case like this, the side of the shell that holds all the syllables has to share one to the other side.
  • In this instance, that would leave us with ble…cŋeṅ as the shell instead, with results I can't completely predict. But it might be flavorful.
  • dragonfly

• Words are mainly written consonant-by-consonant. The root is written with a special glyph of its own. (I think) each word has at most one root glyph.
• It sounds diabolical but when you consider how vowels are shifted around pretty violently in the course of prosodic balancing, it may make words more readable to avoid writing them.
• In this case, the characters written would be: {θi}-B-L-C-Ŋ-N.
• Note that the B, C and Ŋ characters also originated from roots at the start of the etymology. Since they are not the primary root of this word, they have fallen to the wayside and are now written phonetically.
• Find some way of conveying what this would look like.