#troll conlang

[paradoxcase]

@yumantimatter:

👀 link troll conlang?

I think it's variously on this blog in the #troll conlang tag and was developed in more depth on @nowordsforsnow for a while

Uh, it was a Homestuck thing. I feel like I should mention that, since I don't know how obvious that is anymore

[nowordsforsnow]

A long time ago there were some things here that I wanted to fix, but I've honestly forgotten what they all were, and rereading the blog hasn't made anything really stand out to me. But there are a couple things that need to be fixed in the dictionary:

jǔoqǎēvǐqí ("cuttlefish") - we used the wrong diminutive, it was i, should have been í jraojròl ("tree") - compound was head-last âcrâqēuzìrē ("scorpion") - contained illegal consonant cluster

Next, I think we ought to iron out the rē nominalizations, since we have a few of them now. How many of you followers are still here?

[paradoxcase]

Oh hey, in various real life excitements, I missed the last couple Homestuck updates. Pre-retcon Terezi and Vriska reunited, yay! I wish we could have seen them talk to each other, but the flash was pretty nice.

Anyway, so here's what's been going on with me, in condensed version:

I finished all the classes I need for my Master's.

I got a job, which I need to do for a while to complete my Master's.

I moved across the country (to the Massachusetts North Shore) to do the job.

I've been doing the job for two weeks - I've gotten stuff done and it seems pretty great!

Anyway, now that things have settled down a bit, I want to go back to old projects. I probably won't have a lot of time to work on them during weekdays, but I'll probably be able to do stuff on the weekends.

The main thing I want to get done is a conlang development program I've been writing that's available here (if you want to use it and have trouble installing/running/using it, let me know) (unfortunately the Mac version is not up to date - if someone wants to compile it for Mac, let me know and I'll send you the source code for the last version, as opposed to the stuff on github, which is between versions).

I want to continue on with the troll conlang, but first I want to go back through all the posts on nowordsforsnow and do a bit of a revamp, since I think there were some inconsistencies that developed as we went on.

I don't have plans at the moment to do anymore fanfic or continue uncompleted fanfic (or non-fanfic writing projects that no one here knows about), but at some point when the first project reaches version 1.0 or something I might return to it - if Homestuck fandom still exists and/or I get into another fandom enough to want to write about it.

At some point in the past I said that Troll Homestuck Fandom was kind of over because of disillusionment with Homestuck. But I've more or less gotten over that, now, so I might well have an "ask me about THF" day in the future if people are still interested (are people still interested?).

I think that about covers it.

[logodaidalos]

Morphology - Verbs (I)

So, verbs! They're probably going to be the most fun bit of Trollian (temp. name), and where we'll spend the most time.

(For reference, from now on I will be using a simplified transcription scheme, where /ɑ(:)/ will be represented by <a(:)> and /ə(:)/ will be represented by <o(:)>)

Trollian verbs will mostly be built out of either general stems or bare roots. Roots in general will be very verb-y in nature, and verbs would be the primary thing derived from them. Nouns are additional derivations or just re-interpreted verbs.

Some elementary verbs would probably be minimal in size, expressing some rudimentary concepts. A suggested few, for illustrative purposes:

<ʂr> - eat

<ɖʐ> - please

<k'> - shout

<kr> - see

Some other, more complex roots, would embody more complex concepts, such as:

<ne> - accept

<sa:xt> - write

In general, roots cannot stand alone as a verb. Due to grammatical limitations as to what constitutes a sentence, a verb must at least be inflected for the absolutive. I'll tentatively propose the following series of morphemes:

<-ja-> - first person singular absolutive

<-ra-> - second person singular absolutive

<-na-> - third person singular absolutive

Since Trollian verbs are going to become massively inflected, I've been considering a free-er order of affixes, something resembling Greenlandic but not quite like it: affixes do not have a fixed position in the verb and instead are free to move around, causing lots of allomorphy.

Giving a freer order to affixes in the verb gives us travesties such as <ʂrra> and <sa:xtna>, which do not feel like they belong in a language I'm looking to make smooth and flowing. We need to introduce some restrictions and processes.

Primarily, I'm thinking about anaptyctic vowels i.e. vowels inserted to break up unwieldy consonant clusters. As all the vowels in that chart are phonemic (except for the [ə:] which I'll have to elaborate upon later), it is basically a choice of using any of the short vowels. Instead of going with <-o->, I'm thinking of using <-a-> as it's a fairly neutral and pleasant vowel.

Inserting the anaptyctic vowel would be also variable. One could insert it in any desired location, though some clusters might obviously be obligatorily broken up. As it's a morphologically null-valued affix (meaningless morpheme), it could also potentially be infixed into root. This would give verbs such as <sa:xatna> or <sa:xtana>, and <ʂrara>.

A limitation I'd propose is to make all ejectives always come before vowels. This would mean that a hypothetical "he is shouted" would have to be translated as <k'ara> or <rak'a> and definitely neither <k'ra> or <rak'> as it would go against the limitation. Might be arbitrary, but word-final ejectives don't feel good to me.

Another limitation would probably be that clusters with ascending sonorancy (such as <ʂr>, with a low-sonorancy fricative followed by a high-sonorancy approximant) would definitely be disallowed from being syllable codas. This would mean that sonorancy decreases as one moves farther away from the vowel, and it would also disallow a verb such as <jaʂr> because it would have a bad sonorancy order at a coda, and would instead force it to be something like <jaʂra> or <ʂraja>, or even <ʂarja>. 

I think this free-er verb building actually gives a sort of unique feel and flow to the language. If anyone wants to do some work on this, while keeping the morphosyntactic alignment in mind, try translating the following sentences in as many ways as you can, and send here if you've got answers!

"You are pleased"

"I am accepted"

"S/he is seen"

"I am eaten"

—Nexvs

[logodaidalos]

Basics of Grammar - Morphosyntactic Alignment

Morphosyntactic alignment, or in brief MSA, is the relationship between the sole argument of an intransitive sentence (the "subject") and the two arguments of transitive sentences (the "agent" and "patient"). English deals with these arguments by grouping the subject and agent together and differentiating them from the patient (f.e. "I eat" and "I hit him" & "He hits me"), but I've thought that the trolls could do with something called an ergative-absolutive alignment, where the subject and patient are grouped into the "absolutive" case and the agent into the "ergative".

What this implies for the language is, besides the noun marking, some structural differences from f.e. English. As all languages treat the case/category that includes the subject with preference over other categories, it would mean that all normal intransitive sentences in the language would work just like equivalent English passive sentences; what English would express with "I am eaten", the trolls would say as simply "I eat". Transitive sentences would remain the same, mostly.

An additional artefact of this is that we now need to introduce something that would be the mirror opposite image of the English passive: in linguistics it is generally called the "antipassive". This grammatical voice behaves by deleting the absolutive and "promoting" the ergative to the absolutive, just like the normal passive works by deleting the nominative and promoting the accusative to a new nominative ("I killed a bear" > "A bear is killed").

These antipassive sentences would work and mean the same as normal English intransitive sentences, mimicking what happened with the troll intransitive sentences and English passive.

Furthermore, this focus on the absolutive means more focus on the objects of verbs, which is reflected in the grammar: instead of using passive participles (like English "killed" where the action is carried out onto something), we would use antipassive participles (such as the English active participle "killing") more often, and deriving new words would probably go in the opposite direction: where English has a really productive <-er> morpheme that produces verb agents, trolls would use a corresponding and equivalent morpheme that would produce verb patients more often than not. English has the <-ee> morpheme that is infrequent and not quite as productive, and trolls might even lack such an equivalent and would rather use the participle.

Additionally, when leaving out certain arguments of a verb, it would be much more frequent for trolls to leave out the patient rather than the agent/subject. English drops the agent, as seen in the example sentence "I saw the bear and killed it"; trolls would rather say something equivalent to "I saw the bear and I killed" which, while totally wrong in English, would be just fine for them.

This property that the language of the trolls has is called "ergativity" and it is somewhat frequent even among human languages, but almost never to this extent.

—Nexvs

[logodaidalos]

Basics of Grammar - Castes (I)

On Alternia, and in the Alternian Empire, the trolls are the main (if not the only) component. Biologically, trolls are theoretically sorted into one of twelve blood castes, though limebloods, as far as we know, are not an extant caste.

There also exists a dichotomy between seadwellers and landdwelling trolls that permeates all layers of the caste mindset (as seen with Equius' contempt of Eridan). An additional feature of the hemospectrum (the blood caste system) is that there occasionally may be glitches or mutations, which can (as with Karkat) produce trolls of an abnormal blood colour, or give (as with Rufioh) physical mutations of an undesirable kind.

This division into blood castes is so prevalent that it even seeps into Karkat's own speech, even if he is against the hemospectrum. He identifies three different caste types:

When talking to Terezi, he identifies her as a 'blue blood' indirectly via his use of 'your blue blooded vernacular', but when he explains to Dave the nuances of romance in a trashy romance novel he uses the terms 'highblood' and 'lowblood'. Equius additionally uses 'sea dweller' as a label for Eridan's caste:

As Equius identifies Eridan as a sea dweller, but also as a highblood ("his blood is even purplier"), that kind of brings about that there might be multiple caste groupings involved.

I propose the following:

Lowbloods > [Red, Bronze, Yellow, (Lime)]

Midbloods > [Olive, Jade, Teal]

Highbloods > [Teal, Cerulean, Indigo, Purlple, Violet, Tyrian]

Additionally, I propose another division level that deals specifically with more general groupings by colour terms which excludes the lowbloods whose blood colours differ significantly:

Greenbloods > [(Lime), Olive, Jade, Teal]

Bluebloods > [Teal, Cerulean, Indigo, Purple]

Seadwellers > [Violet, Tyrian]

In both groupings, teal acts as a sort of a boundary, meaning that it can be used as both a midblood/greenblood colour or as a highblood/blueblood colour when appropriate in context.

Further, due to the massive impact on troll society and culture that the hemospectrum and castes have, a formality system based both on social status and blood colour would definitely be present. In form and function this system would likely resemble the formality system of Japanese, with honorific speech and extensive formality in the grammar itself.

On the other hand, I don't think that castes would be marked on the nouns themselves (which is what the conlang on nowordsforsnow did), but rather it seems better to mark the nouns with a sort of an animacy system. In that case, nouns would belong to one of three classes: terrestrial (land-based), aquatic (sea-based) and inanimate. Twelve noun classes might be a bit overkill.

—Nexvs

[logodaidalos]

Phoneme Inventory

The first step towards making a conlang would probably be figuring out what contrastive sounds make up its phoneme inventory. The phonemes are the theoretical contrastive sound units that function as building blocks of morphemes, not the actual sounds that a mouth produces; the real sounds are phones and are dealt with under allophony and general phonetics.

For the purposes of inspiration gathering, the Logodaidalos always turns towards canon sources, but these are pretty non-existent (at the present time), so we're forced to be a bit more innovative. What struck me as a good parallel was the caste system of Alternia versus the caste systems of some human societies -- in this case, Tulu castes, which are arguably more complex than Alternian ones --  and so it might be a good parallel to draw between the phonology of a casted people and of Alternian trolls.

The phonology of Tulu is very Indian in its nature, as well as characteristically Dravidian. It contrasts retroflex consonants with denti-alveolar ones, and has a palatal series that is undergoing a shift to a postalveolar affricate series. It has a voicing contrast, but no aspiration or breathiness contrast present in a lot of other Indian languages. It also has a length contrast on six vowel qualities and one back unrounded vowel.

More inspiration the Logodaidalos found in the Caucasus. The Caucasus Sprachraum (language zone) is made up of multiple language families that have acquired characteristics of one another and developed some share features, namely ejective consonants (Ossetian, an Iranian language, has developed them from being in the zone) as well as a very high level of agglutination bordering on polysynthesis (as exemplified by Georgian and Tsez).

An additional, now thematic, inspiration the Logodaidalos found in the Nāhua people, of whom the Aztecs are/were the most notable members. Classical Nāhuatl is extremely polysynthetic, but with a relatively simple phonology that doesn't distinguish voicing in consonants and with four vowel qualities. It is also characterised by its prevalent use of the lateral affricate, /t͡ɬ/ tl, as a non-possessedness marker, and by widespread reduplication.

Mixing and matching some notable features of these interesting and different languages, we arrive at something that has properties of each group:

This table has many features many languages above described have, and is made up of some elements that are more-less unremarkable on their own but together form an unique blend. It has a whole set of ejectives, from all three MOA (method of articulation) classes. It's quite interesting in that it distinguishes ejective and tenuis affricates and fricatives, along with having gemination only on non-sonorant obstruents.

For vowels it is a bit more tame than what can be found on the Caucasus, incorporating some Dravidian and Nāhuatl elements:

The schwa may act as a sort of a reduced vowel that is inserted in predetermined positions and that allophonically lengthens under stress. The exact mechanism of its action is as-of-yet undetermined.

Stress in the language would probably be non-phonemic and uncontrastive, most likely determined by vowel length and syllable position, with a tendency to fall on the penultimate or antepenultimate.

Allophony for the language has yet to be worked out, but if anyone is willing to think about it, feel free to write on it and notify the Logodaidalos!

--Nexvs

[logodaidalos]

A Summary of Previous Attempts

This is definitely not the first such troll conlang attempt. I know of only one that panned out significantly (i.e. beyond the phonetics jab), and that is over on nowordsforsnow. The blog has some extensive morphology laid out, along with basic grammatical concepts (blood-castes dictating noun classes, for one) and some varied and topical lexica. The conlang there builds on the phonetics I wrote about in a previous post. That blog has quite a following for this sort of thing (which let's hope we'll manage here!), but, more importantly, turns up first when "troll conlang" is searched on tumblr. The blog has been fairly inactive for more than a year and a half.

Looking tags up like "homestuck linguistics" gives one additional post, and looking up "homestuck conlang" gives only the Logodaidalos' posts. So far it seems like a moderately empty field with lots of creative wiggling room. Some inspiration might be found in the nowordsforsnow's conlang, of course, but most of the ideas here I'm hoping will be original to as great an extent as I will be able to muster.

There is probably a lot to do, starting from an organisational and logistics point of view (what topic needs the most time allocated to it?) and ending with the questions of realisation (how to sort and present a dictionary f.e.), but it can be done. If anyone knows of any other troll conlang attempt, send a message and it'll be taken a look at!

--Nexvs