Dunsish Scratchpad

[duns]

Hi all. I've been sketching a language in my mind for awhile, several years in fact, and with the pandemic I've decided now's the perfect time to actually commit it to writing. The language is just called Dunsish for now, but I'm not sure if that will stay. I was going to do a full sketch but I think instead I'm going to work out the bits of the language on here piecemeal. In general, I am not going for absolute naturalism here. I want to actually use the language, and I want it to be fun to use, and my idea of fun is not memorizing paradigm tables, so there's likely to be a higher degree of regularity than one might find in natlangs or more naturalistic conlangs. Anyway, Let's start with phonology

Phonology

I'm going for something here that feels a bit Germanic, perhaps with a touch of Finnic/Sami. I've also added a series of ejective stops, because I'm gotten good at pronouncing them from learning Georgian, and I like the sound. I also want to accommodate certain Kartvelian loanwords.

Consonants

	Labial	Dental	Alveolar	Palatal	Velar	Glottal
Nasal	ṃ m		ṇ n
Stop	p b		t d		k g
Ejective	p'		t'		k'
Sibilant			s
Fricative	f v	þ ð		c		h ǥ
Aproximant				_ j	ƕ ƿ
Liquid			ḷ l
Trill			ṛ r

All consonants have their IPA values with the following exceptions:

• <ṃ> = /m̥/
• <ṇ> = /n̥/
• <c> = /ç/, patterns like the voiced counterpart of <j>.
• <ǥ> = /ɦ/, in actuality some kind of voiced pharyngealized sound, though I'm not sure what yet. Nevertheless, it patterns like the voiced counterpart of <h>.
• <ƕ> = /ʍ/
• <ƿ> = /w/
• <ḷ> = /l̥/, generally realized with some (light) frication, so [ɬ] in narrow transcription.
• <ṛ> = /r̥/

There are some open questions here: for instance, are the alveolar stops apical or laminal, for instance? (I am leaning towards the latter). These will be filled in latter, I suppose. Geminated consonants are contrastive with non-geminates but only occur word-medially and word-finally. Geminates only occur as a result of compounding or inflectional suffixes.

Vowels

I like rich vowel inventories, so there's a lot here. Vowel height factors into the morphophonology of Dunsish, so this table records first and foremost what its vowel phonemes pattern as rather than their exact realization.

	Front	Back
Close	i y	u
Close-mid	ȝ w	o
Open-mid	e ø	ǫ
Open	æ œ	a

All vowels have roughly their IPA values except for the following:

• <ȝ> = /ɪ/
• <w> = /ʏ/, I'm aware this orthography is somewhat eccentric given that I'm using <ƿ> for /w/, but my rationale here is that basically no fonts I have will render the acute accent over the wynn correctly, which is important for reasons which will be given below.
• <e> is mid, so it's more like [e̞], but /e/ works well enough in broad transcription.
• <ǫ> = /ɔ/
• <œ> actually does have its IPA value but it's worth just mentioning that it's of course open-mid, not open. I have it in the final row because it morphophonologically patterns like <æ> and <a>.
• <a> = /ɑ/

There are five diphthongs: au, ǫu, ei, æi, øy.

Vowels come in three different contrastive lengths. However, overlong vowels can only appear in in the middle of roots. If I were of the diachronic persuasion there'd probably be a straightforward way to explain this. Unlike Ancient Greek but similar to Old English, diphthongs can be short, long, or (in the case of Dunsish) overlong.

Syllable Structure

I have an idea of what syllable structure should look like but I haven't quite formalized it yet, it's probably something like: sCCjVNCC

There are probably restrictions on where short vowels can occur. I've found myself compensatorily lengthing short vowels before unvoiced liquids, so this should probably end up codified in the language itself.

Prosody

To the extent that isochrony "exists", Dunsish is mora-timed. I don't know much about prosody in general, so this might be complete nonsense, but my current idea is that (of course), short, long, and overlong vowels are mono-, bi-, and trimoraic, respectively. Moreover, syllable codas with a geminate consonant, nasal, lateral, or rhotic also contribute a mora. Thus under this view, a syllable could be maximally tetramoraic. I'm not sure if such a system is attested, I'd need to do more research here.

Tone/Pitch Accent

I'm in general skeptical of the concept of "pitch accent", however we can use it without too much violence towards linguistic precision to refer to Dunsish's rather meager tone system. In essence, in every word (root, more precisely, since compounds maintain pitch on all roots that compose it) zero or one vocalic mora carry a high tone. This is in contrast to to e.g. Swedish where there is a word-level tone contour, but similar to Japanese or Ancient Greek. Certain derivational infixes or inflectional suffixes can cause the tone of the right-most root (if in a compound) to shift. Pitch accent is indicated orthographically by an acute accent. Since Dunsish has contrastive vowel length, this results in nine different tone contours:

• short low a
• short high á
• long low aa
• long falling áa
• long rising aá
• overlong low aaa
• overlong falling áaa
• overlong rising falling aáa
• overlong rising aaá

Phonological rules/processes

This is somewhat unorganized right now. While some of these rules are things I prescribed, some are things I codified that I found myself doing naturally.

• /l/ is velarized after /ɑ/. /hɑl/ is thus [hɑɫ]. This I found myself doing to distinguish between e.g. /hæl/ and /hɑl/.
• overlong <eee> and <æææ> diphthongize to [eːːo] and [æːːɑ]. These are similar to Old English <eo> and <ea> (but overlong, of course).
• non-geminate <k> and <g> lenite to [x] and [ɣ] intervocalically, respectively (and possible other circumstances, I'm just sure of this one, possibly before /t/).
• unvoiced stops "pre-aspirate" postvocalically, similar to Icelandic or Faroese.
• /k/ and /k'/ are realized as [q] and [q'] before /ǫ/ and /ɑ/.
• /k/, /k'/, and /g/ are realized as [c], [c'], and [ɟ] before <i y ȝ w>
• /n̥/ and /n/ are realized as /ɲ/ /ɲ̥/ before <i y ȝ w> or before [c c' ɟ]
• /l/ is realized as [ʎ] before <i y ȝ w> or before [c c' ɟ]
• word finally, a vowel preceding the voiceless nasals is nasalized.

Morphophonology

This will only be the briefest of sketches, to be filled in later.

All roots consist of a single closed syllable. In essence, there are three kinds of derivational morphophonological processes which produce concrete stems from abstract roots:

1. Initial consonant mutation
2. Root vowel apophony
3. Derivational infixing

I have not worked out exactly what mutation looks like, voiced/unvoiced would be obvious, but I'm not sure if I want to put that much functional load on a simple voicing contrast. I am also not sure what purpose this will serve yet. One idea is that, since <s> has no voiced counterpart, s- could serve as a prefix which blocks mutation.

Root vowel apophony is broadly used to transform roots from one part of speech to another (this isn't entirely accurate, but for reasons I'll go into in a later post about the way the verbal system works). So noun→verb, noun → adjective. The root vowel is abstract and only specifies height, its concrete realization in the stem depends on part of speech and is either front unrounded, front rounded, or back. I'm currently unsure how diphthongs factor in here. Root vowel length is inherent to the root, and no inflections modify it.

Infixing is used for less coarse derivational processes, so think dimunitives, causatives, etc. There is probably some kind of height harmony here although I'm not certain on that.

[Vardelm]

Welcome to the ZBB! Enjoy the customary, delicious greeting:

Hi all. I've been sketching a language in my mind for awhile, several years in fact, and with the pandemic I've decided now's the perfect time to actually commit it to writing.

There's nothing like forcing yourself to make a post & present the ideas to other people to make a conlang really solidify, is there?

I was going to do a full sketch but I think instead I'm going to work out the bits of the language on here piecemeal.

I've had good progress lately since the "scratchpad" idea helps get past the urge to make it "perfect" before presenting.

In general, I am not going for absolute naturalism here.

It's helpful to know someone's goals for the language! Still, I think feedback will lean towards naturalism, just because comparing to real languages is about the only yardstick there is for conlangs. "Non-natural" conlangs are perfectly valid exercises, but the more one gets away from naturalism the harder it may be to give helpful feedback.


All that said, what you have here looks good so far. There's some phones in there that I don't see used often, which adds some interest. I'll be curious to see what you come up with for consonant mutations since I'm working on a language with CM as well. Finally, I like the limits on the use of infixes. Nice touch.

[duns]

Welcome to the ZBB! Enjoy the customary, delicious greeting:

Thanks!

In general, I am not going for absolute naturalism here.

It's helpful to know someone's goals for the language! Still, I think feedback will lean towards naturalism, just because comparing to real languages is about the only yardstick there is for conlangs. "Non-natural" conlangs are perfectly valid exercises, but the more one gets away from naturalism the harder it may be to give helpful feedback.

Yeah, I probably should have clarified what I meant here a little bit better. I am definitely aiming for a kind of aesthetic naturalism, inasmuch Dunsish is definitely not an engelang. Nevertheless, some of richness and complexity one would expect to find in a natlang, demanding thousand page grammars, is something I'm deliberately eschewing since I want to actually learn and use the language. With that in mind, I would definitely appreciate feedback on features or misfeatures that feel contrived. With that being said, I actually have some features of the language planned that I'm fairly certain aren't present in any known natural language, but I think that something being natural iff it's extant in the literature is a bit too narrow an understanding.

All that said, what you have here looks good so far. There's some phones in there that I don't see used often, which adds some interest. I'll be curious to see what you come up with for consonant mutations since I'm working on a language with CM as well. Finally, I like the limits on the use of infixes. Nice touch.

Thank you, I appreciate the feedback! How do mutations in your language work (both in terms of their function, and the phonological process)? I've had a couple ideas myself, none of which I liked. The first was some kind of privative function (think mal- in Esperanto) but in my opinion this is clearly an awful idea, especially if the mutation is merely a voicing contrast. That's way too much function load. The other was that mutating the root would result in a new stem which functioned as a "metaphorical" version of the original. I think that's a bit too twee.

[duns]

This will be a short post.

Gender System

There are several different categorization schemata that nominal roots/stems participate in, but right I'm just going to focus on the gender system. In essence, there are 5 different genders which correspond to the 5 (western) classical elements: earth, water, air, fire, and ether. The gender of roots is in general less arbitrary than the typical Indo-European scheme (that is to say, more or less completely save for nouns which actually have a sex) but more so than e.g. the system present in the Dravidian languages. Rarely, derivation can produce a stem with a different gender than the root that it is derived from, but this is unusual. The descriptions below are definitely not exhaustive.

Earth

This is the most common gender and often ends up including nouns which do not neatly fit in the others.

Water

The water gender includes aquatic animals, liquids, many abstract nouns relating to biology,

Air

The air gender includes birds, most insects, gases, and things related to the sky such as clouds, but also many kinds of celestial bodies.

Fire

This is probably the rarest gender. Anything related to fire, heat, destruction, or death is included in this gender. Notably, body parts, even those of living people are included here.

Ether

The ether gender behaves somewhat differently from the rest. It included humans but also abstract ideas such as mathematical objects, as well as supernatural beings. I've not yet fully worked out how this one works.

Morphology

Each of the non-ether genders has a thematic vowel. All case/number suffixes for a given noun include the thematical vowel.

Gender	Thematic vowel
Earth	i
Water	u
Air	a
Fire	e

[duns]

Verbs

I want to talk a bit about the verbal system of Dunsish. Like Basque, colloquial Welsh (as I recall), Udi, and others: Dunsish has a closed class of light verbs which carry aspectual and modal inflections (notably not tense, however). There are approximately 30 or so light verbs, most of which are very semantically light and generic. However, include in this list are verbs of motion which can be relatively detailed.

We'll return to light verbs in a second, but we need to talk about verbs proper, first. "Verbs" in Dunsish (besides the aforementioned light verbs) are nouns which refer to the activity itself. There are two things which demonstrate that these are "truly" nouns, and not finite verbs. First of all, Dunsish contains a special case, the verbal case, which the action noun inflects for. So for instance, one would say something like:

do-IND run-VERB.SING 1s.SING.NOM

"I run"

I've removed all agreement markers and whatever else here to focus on what's relevant. The verbal case is a true case like any other, and for instance, "run" could have been inflected in the plural to say that I ran multiple times. Moreover, the light verb agrees with the noun in the verbal case in both gender and number. What English uses adverbs for, Dunsish simply has adjectives, which when modifying a noun in the verbal case, achieves the same effect.

The second bit of evidence that Dunsish "verbs" are true nouns is that they participate in the same derivational processes as all other nouns. Conceptually, all nouns in Dunsish have, among others, two properties: they are ±extended and ±discrete. Neither property implies the others. Electrons, for instance are -extended but +discrete: they take up no space but constitute countable individuals. Now, in nouns referring to actions, these properties end up corresponding to aktionsart. In Vendler's terminology, discreteness corresponds to telicity, and and extension corresponds to durativity. We can, for instance use the same derivational infixes to transform the concept of "running" into "a run".

Now, to bring this back to light verbs, the lexical aspect/aktionsart of a particular verb-like noun partially determines which light verbs it can be used with. Some have different meanings when used with different light verbs.

I think that's all I want to say for now!

[Vardelm]

Yeah, I probably should have clarified what I meant here a little bit better. I am definitely aiming for a kind of aesthetic naturalism, inasmuch Dunsish is definitely not an engelang. Nevertheless, some of richness and complexity one would expect to find in a natlang, demanding thousand page grammars, is something I'm deliberately eschewing since I want to actually learn and use the language. With that in mind, I would definitely appreciate feedback on features or misfeatures that feel contrived. With that being said, I actually have some features of the language planned that I'm fairly certain aren't present in any known natural language, but I think that something being natural iff it's extant in the literature is a bit too narrow an understanding.

Fair enough!

How do mutations in your language work (both in terms of their function, and the phonological process)?

So far, I don't know exactly, but word-initial will work much like Celtic languages, while word-final similar to Uralic. For the starting (proto) language that I'm working on now, it will just be phonological, but descendant languages will use only mutation for noun case & verb voice.

I've had a couple ideas myself, none of which I liked. The first was some kind of privative function (think mal- in Esperanto) but in my opinion this is clearly an awful idea, especially if the mutation is merely a voicing contrast. That's way too much function load. The other was that mutating the root would result in a new stem which functioned as a "metaphorical" version of the original. I think that's a bit too twee.

I don't think that's necessarily awful. Voicing (& other contrasts) seem to be used for a lot of different uses. You might be underestimating the functional load it can carry. At the same time, I tend to do the same.

"Twee" might depend on the context: IE, what else is going on in the 'lang.

[tlacamazatl]

When you get a chance, I'd enjoy you expanding on the gender system, as I find it the most interesting aspect you've presented thus far.

[Vardelm]

I'd enjoy you expanding on the gender system

Agreed. The use of the 4 (5?) elements could be "twee" (as you put it), but expanding them into concrete noun genders/classes makes it feel fairly natural, as though this is just the way speakers view their world, instead of being some meta-creative statement that this is the structure of reality.

The second bit of evidence that Dunsish "verbs" are true nouns is that they participate in the same derivational processes as all other nouns. Conceptually, all nouns in Dunsish have, among others, two properties: they are ±extended and ±discrete. Neither property implies the others. Electrons, for instance are -extended but +discrete: they take up no space but constitute countable individuals. Now, in nouns referring to actions, these properties end up corresponding to aktionsart. In Vendler's terminology, discreteness corresponds to telicity, and and extension corresponds to durativity. We can, for instance use the same derivational infixes to transform the concept of "running" into "a run".

I rather like this bit, too. Extending telicity to nouns is clever (or, at least, I haven't seen it before). Seeing how the light verbs will interact will be fun.