During the past three months, I’ve been struggling to devise a common ancestor for Ryka and Quriil. Reverse-engineering a plausible proto language for two conlangs that already exist is always a difficult task, although fortunately, Quriil is not much more than a draft yet, so it is still flexible enough to receive minor changes to its phonology in the process, and it does not yet have any vocabulary except for a few character names.

Let’s first have a look at what we already have for the two languages. What do their consonant and vowel inventories look like? What do they already have in common? Are there any oddities that we might find a good explanation for via the proto language?

Ryka’s phonology

For Ryka, I’ll have a look at the phonology of Old Ryka (or “Rykaic”), the common ancestor of the contemporary Rykaic languages. It is very close to Capital Ryka though.

	Bilabial	Dental	Alveolar	Velar	Uvular?	Glottal
Stop	p / b p / b	t / d t / d		k / g k / g		q ʔ
Fricative	ph ɸ	th θ	rh r̥~ʂ	kh x	(‘) ʁ~ʕ	h h
Trill			r r
Lateral			l l

Old Ryka’s consonant inventory is quite systematic due to the consonant mode system. It has five main places of articulation: Bilabial, dental, alveolar, velar and uvular/glottal. For each of these places of articulation except for the alveolars, there is a voiceless stop, a voiced stop (fricative in case of uvular/glottal), and a voiceless fricative. In the alveolar column, we have a voiced and a voiceless trill as well as the voiced lateral approximant. In Ryka’s own phonological theory, they align with the stops and fricative of the other places though. There are no nasal consonants.

	Front	Central	Back
High	y ɨ ~ i ~ i̹ ~ y		u u
Mid	e ɛ		o ɔ
Low		a a
Vocalic consonant		n m̩~n̩~ŋ̩

Old Ryka has a typical five-vowel system consisting of /a/, /ɛ/, /i/, /ɔ/, and /u/. It also has a vocalic nasal consonant that can act as the sole nucleus of a syllable, as in qnk [ʔŋ̩k] “stone”. The high front vowel /i/ acts a little weirdly, being slightly or fully rounded under certain conditions, or moving a little bit to the back, being pronounced more like [ɨ]. For this reason it is transliterated as y, rather than i.

	a	e	o	u	y
a	/	aej aɛ̯	aow aɔ̯	auw au̯	ayj ai̯
e	jea ɛ̯a	/	eow ɛɔ̯	euw ɛu̯	eyj ɛi̯
o	woa ɔ̯a	oej ɔɛ̯	/	ouw ɔu̯	oyj ɔy̯
u	wua u̯a	wue u̯ɛ	wuo u̯ɔ	/	uyj uy̯
y	jya i̯a	jye i̯ɛ	jyo y̯ɔ	jyu y̯u	/

Old Ryka has a lot of diphthongs, any combination of two vowels is possible. In terms of prominence, there is a hierarchy /a > ɔ, ɛ > u > i (> n)/. Where /ɛ/ and /ɔ/ tie, the first vowel is prominent. In Old Ryka, pairs like /ɛɔ̯/ and /ɛu̯/ or /ɔ̯a/ and /u̯a/ were actually contrastive, while /ɛ̯/, /i̯/ and /ɔ̯/, /u̯/ have merged in Capital Ryka, but are still retained in the orthography.

The syllable structure is CV(C). An empty onset is not allowed. There can be word-medial clusters of two consonants, but they need to agree with respect to “consonant mode” (roughly the manner of articulation). This is a very Rykaic thing though and can be an innovation there. Stress is assigned according to predictable rules (on the first diphthong or before geminate consonants, otherwise on the first syllable), so it needs no explanation from the proto language.

Some preliminary thoughts on Ryka:

• The consonant system of the proto language does not yet have to be this systematic – the pressure for mode assimilation should be sufficient to put the consonants into their proper places after Ryka split from Quriil. The whole consonant mode system is supposed to be a Rykaic feature.
• The alveolar consonants are predestined to have evolved from proper stops – alveolar or retroflex, for instance. That can help in explaining why they are treated equivalent to the other stops and fricatives. However, we can also introduce them as trills and laterals and say that the consonant mode system made them fit in.
• I can imagine the glottal/uvular consonants to have shifted from another place of articulation, uvular-only, for instance. The glottal stop is already transliterated as q, so that would fit into the whole narrative. (Not that orthography should be a criterion in this, but it feels natural.)
• Not having any nasal consonants (disregarding the vocalic nasal) is quite marked, so our proto language should probably have them. This means that Ryka needs to have lost or transformed them somehow.
• We need to be extra careful with the vocalic nasal, as it is a rather unusual feature. If it is only introduced in Ryka, it needs a very good explanation. If that’s not possible, we should already introduce it in the proto language. It’s much easier to explain why all the other languages lost it than why Ryka innovated it – also considering that the modern Rykaic languages have also partially degraded/lost it.
• Creating this proto language is the perfect opportunity to find an explanation for Ryka’s quirky /i/. I’ve long entertained the idea that earlier forms of Ryka used to have an additional consonant /ɨ/ that merged with /i/, and I think we should pursue this idea here.
• Reminder to not forget the diphthongs.

Quriil’s phonology (so far)

Quriil is more a collection of ideas than a proper conlang yet, so we still have some freedom to change things here. This will definitely help in creating a good proto language. Here’s what I came up with so far:

	Labial	Alveolar	Lateral	Retroflex	Palatal	Velar	Uvular
Asp. Stop	ph pʰ	th tʰ		tth ʈʰ		kh kʰ	qh qʰ
Plain Stop	p / b p / b	t / d t / d		tt ʈ		k k	q q
Affricate			tl t͡ɬ
Fricative	f / v f / v	s / z ʃ / ʒ	lh ɬ	sh / zh ʂ / ʐ	sj / zj ɕ / ʑ		x χ
Nasal	m m	n n		nn ɳ	nj ɲ
Trill		r r
Glide			l ɫ	rr ɻ	j j

Quriil has a lot of consonants. There are six places of articulation: Labial, Alveolar, Retroflex, Palatal, Velar and Uvular. Most of these have stops, both plain and aspirated, fricatives, both voiced and voiceless, and nasals, but there are gaps. In the labial and alveolar column, there is even a threefold distinction in stops (voiceless aspirated, plain voiceless, voiced). There are no palatal stops and no velar fricatives (looking at this now, it almost looks like they belong together, and everything velar except for the stops was palatalized, although I cannot remember doing this intentionally). There is no voiced uvular fricative and also no uvular nasal. In addition, there are two rhotics, an alveolar trill and a retroflex glide, and three lateral consonants: Affricate, fricative, and approximant.

There are a few groups of consonants that I would like to keep:

• The three laterals, because I really want to have a language with a lateral affricate and fricative, and it fits Quriil’s general aesthetic.
• The uvular stops and fricative, because they fit Quriil’s general aesthetic and hint at the Aleut inspiration.
• The retroflex consonants (at least some), as a tribute to my beloved Dravidian languages.
• The aspirated/plain distinction in stops (at least some). The voiced stops can go, if necessary.

	Front	Central	Back
High	y / i / ii ɪ̆ / i / iː		w / u / uu ʊ̆ / u / uː
Mid		e ə̆
Low		a / aa a / aː

Quriil’s vowel system is the phonological feature around which I built everything else, so it’s obviously a keeper. Quriil has a threefold vowel length distinction: Long vowel, short vowels, and super-short centralized vowels. Disregarding this, there are only three distinct vowel qualities: /a/, /i/, and /u/. The vowel’s actual pronunciation changes heavily though, depending on the surrounding consonants. The short vowels will also often be dipthongized when stressed. The super-short vowels may vocalize and merge into neighboring sonorants. Quriil has no real diphthongs apart from its dipthongized stressed short vowels.

I don’t have strong feelings about the syllable structure yet, but the names I came up with so far only have very few consonant clusters, as heavy consonant clusters in combination with the super-short vowels (see below) would render the language unpronouncable for me. Haven’t really thought about the stress pattern – on the few words I have it’s mostly on the long vowels or on the first syllable, similar to Ryka.

A comparison of Ryka and Quriil

Consonants

When it comes to the consonants, we can see that Ryka and Quriil have a very similar (equal when we merge Quriil’s palatals and velars) number of places of articulation that map nicely onto each other:

• Ryka’s Bilabials vs. Quriil’s Labials
• Ryka’s Dentals vs. Quriil’s Alveolars
• Ryka’s Alveolar’s vs. Quriil’s Retroflexes
• Ryka’s Velars vs. Quriil’s Velars
• Ryka’s Glottals vs. Quriil’s Uvulars

Quriil has a much wider array of manners of articulation, however, and this poses quite the challenge. The proto language either needs to have a similarly rich consonant inventory, with Ryka losing some of the distinctions, or Quriil needs to innovate some features, like voicedness, fricatives or nasals. For this, however, we need a condition that triggers this change in some consonants, but not in others, and the condition needs to survive somewhere, but not in Quriil, so that the new consonant distinctions become phonemic, but the evolution can still be reconstructed.

One idea that I’ve entertained for a while: A simple high-low tone distinction on the vowels could trigger assimilation on the consonants. There is an association between high tones and voiceless consonants, and low tones and voiced consonants in many languages, and there are known cases of tones voicing or devoicing consonants in a language¹Maddieson, Ian (1978). Tone effects on consonants. Journal of Phonetics, 6(4), 327-343.. So a tone distinction in Proto-Ryka-Quriil could trigger voicing for Quriil, but not for Ryka, and then get lost in both languages. However, since the tone is not retained in any child language, this will make the voicing seem to appear out of nowhere. If Ryka and Quriil were real languages, a comparative linguist would have a hard time explaining the voiced consonants in Quriil, as they would just magically be there. I find this rather unsatisfying, so I abandoned the idea.

Of course, we can also just choose to reduce the size of Quriil’s consonant inventory and get rid of some stops and fricatives.

The syllable structure of the two languages seems to be quite similar, so no issues there.

Vowels

When it comes to the basic vowel qualities, there’s no fancy stuff happening in either of the two languages. They share /a/, /i/, and /u/, and while Ryka additionally has /e/, /o/, these two should be easy to introduce. Each of the two languages has one feature, however, that is more challenging to explain: Ryka’s vocalic nasal and Quriil’s three vowel lengths.

Ryka has absolutely no vowel length distinction. It does have a lot of diphthongs, but these cannot be plausibly derived from long vowels, as there are just too many combinations of them! We definitely need to have the diphthongs in the proto language, otherwise, the distinction between e.g. /eo/ and /eu/ is simply not justifiable. As Quriil does not have any real diphthongs, they could actually provide a source for the long vowels. But where can we get the super-short vowels from?

One possibility would be to align them with the vocalic nasal. The proto language could have a vocalic consonant (not necessarily a nasal) that became a real super-short vowel in Quriil. However, that would mean that we only have one super-short vowel, unless we establish three vocalic consonants in Proto-Ryka-Quriil… Alternatively, we could introduce the central vowel [ɨ] into the proto language to become Quriil’s super-short vowel and Ryka’s quirky /i/. This would also only serve us a single super-short vowel though.

When it comes to the vocalic nasal, one idea I’ve toyed around with for a while was to make the proto language have pre-nasalized stops (/mb/, /nd/, etc.). In Quriil, these could have evolved into simple nasal consonants, and in Ryka, they could have become the vocalic nasal plus the voiced stops. E.g. a word like /tanda/ would become /tana/ in Quriil and /t̪n̪̩d̪a/ in Ryka, or a word like /ŋgi/ would become Quriil /ni/ and Ryka /ʔŋ̩gi/. However, that would impose several restrictions on the vocalic nasal, that we would need to overcome with more complicated sound changes, for instance that it can only stand before a voiced stop or that it only occurs in bisyllabic words.

The best solution for the vocalic nasal still seems to have a vocalic consonant already in the proto language.

Adding a new player: Konoic

While playing around with all the different solutions, I had another thought that complicated things even further: What if… the Konoic languages were also related to Ryka and Quriil, placing all my conlangs in the same language family? The language tree would look like this:

• Proto-Everything (← this is what we want to reconstruct now)
  • Proto-Ryka-Quriil (← this is what we tried to reconstruct so far)
    • Old Ryka
      • Capital Ryka
      • Balconian Ric
      • other Rykaic languages
    • Quriil
  • Kono
    • Old Nyirvón
      • Nyirvón
      • Siidde
    • Asiul

Having yet another branch of languages in the game to which the proto language needs to be fit is both problematic and beneficial. It is a problem because there are more players we need to please, but it also opens new opportunities for weird sound changes, as there are more languages that can show different reflexes of them to make them more plausible. If the Konoic languages were tonal, for example, using tone to generate Quriil’s consonant inventory would be a no-brainer. They are not, unfortunately, so they don’t help with that, but they gave me some fresh ideas for the proto vowel inventory.

A short introduction to the Konoic languages: Asiul is… my oldest conlang, but also the worst, and it’s needs to be heavily reworked to the point where I’ve actually accepted that I need to completely reinvent it. So I’m open to dropping basically all of its previous features. When reconstructing the Proto-Everything language, we don’t need to be considerate of Asiul. Nyirvón, on the other hand, is a little less flexible. It’s also a draft, similar to Quriil, and while I was once at the point where I had the phonology and even a full set of sound changes from Kono in place for both Asiul and Nyirvón, ready to build vocabulary and construct the grammar, I’m now unsatisfied with what I did back then, in particular for Asiul. So Nyirvón is also open for renovations.

However: Nyirvón is a tribute to Hungarian and is supposed to have a similar sound inventory and phonotactics. In particular, it needs Hungarian’s vowels and the vowel harmony system. This is a stark contrast to the simplistic vowel inventories of Ryka and Quriil. But it actually turned out to offer solutions for some of my struggles!

A proto language with front-back vowel opposition

I began thinking about a proto vowel inventory with a systematic front-back or advanced-retracted tongue root opposition, consisting of e.g. the vowels /a/, /i/, /u/ with both a front (e.g. [æ], [i], [y]) and back (e.g. [ɑ], [ɨ], [u]) realization. We could then start out with a simplistic consonant inventory with only three places of articulation, labial, coronal and radical, and have the nearby vowel’s frontness feature determine their actual pronunciation. The coronal consonant could be /t/ before front vowels and /ʈ/ before back vowels, for instance, and the radical consonant /k/ before front vowels and /q/ before back vowels. In Ryka and Quriil, the front-back distinction of the vowels would then be lost and the assimilated consonants could become actual phonemes. With Kono keeping the front-back vowel contrast, we would end up with a first set of transparent sound changes that are both plausible and drastic enough to obscure the genetic relationship between Ryka-Quriil and Kono.

Giving justice to the inventors’ physiology

Rasvrisuam are shapeshifters. They can learn to take the form of any species they observe, but there are forms that are more practical than others. Nowadays, the human form is prevalent, but before that the most commonly used form was that of the native harpy, an Archeopteryx-like six-limbed feathered reptile. It is likely that the inventors and initial speakers of Proto-Rykaic also mostly, if not exclusively, used the harpy form. That of course alters the range of sounds they were able to use.

I’m not an expert on animal anatomy, but the simplest restriction that comes to mind is the lack of lips, and therefore the unability to produce labial consonants and rounded vowels. Therefore, I decided to cross these from the proto inventory. They can be introduced later in languages such as Ryka once the speakers are primarily in human form, but perhaps some descendant languages lack these sounds until today?

The phoneme inventory of Proto-Rykaic (Proto-Everything)

	Alveolar	Palatal	Retroflex	Velar
Asp. Stop	th tʰ		ṭh ʈʰ	kh kʰ
Plain Stop	t t		ṭ ʈ	k k
Affricate		ch t͡ɕ
Nasal	n n	ñ ɲ	ṇ ɳ	ng ŋ
Approximant		j j
Trill	r r

Since I decided to throw out the labial sounds, I introduced an alveolar-retroflex contrast instead of a labial-coronal one. We then have three main places of articulation: Alveolar, retroflex and velar. These consonants come as aspirated and plain voiceless plosives and as nasals. To introduce a little asymmetry, I added the alveolar trill and some palatal consonants: An affricate, a nasal and the palatal glide /j/. All of these are rather common sounds, so they contribute to the naturalness of the consonant inventory despite the lack of labials, and they can easily be sound changed away in languages that don’t need them, but may serve as a source for additional consonants in others.

Proto-Rykaic probably also had geminate consonants, as many of its potential children have it (at least Ryka and Nyirvón, possibly also Asiul).

	Front	Front-central	Back-central	Back
High	ï / ïï i / iː	i / ii i̠~ɨ̟ / i̠ː~ɨ̟ː	ü / üü ɨ̠~ɯ̟ / ɨ̠ː~ɯ̟ː	u / uu ɯ / ɯː
Low	ä / ää æ~a / æː~aː			a / aa ɑ / ɑː
Vocalic consonants	l̈ ʎ̩			l ɫ̩

I decided that Proto-Rykaic has three pairs of full vowels and one pair of vocalic consonants. All of them are unrounded. We have two low vowels, front /æ/ and back /ɑ/. There are four high vowels; I assume that due to the longer oral cavity of harpies, these four vowels would be easily distinguishable. They form two front-back pairs: /i/ vs. retracted /i̠/ and /ɯ/ vs. advanced /ɯ̟/. Finally, the vocalic consonant is /l/, and it might also have had a front-back contrast between palatal /ʎ̩/ and velarized /ɫ̩/. All full vowels can be short or long; there is no length distinction in the vocalic consonant.

	a	i	u	aa	ii	uu
a	/	ai / äï ɑɨ̯ / æi̯	au / äü ɑɯ̯ / æɨ̯	/	/	/
i	ia / ïä ɨ̯ɑ / i̯æ	/	iu / ïü i̠ɯ̯ / iɨ̯	iaa / ïää ɨ̯ɑː / i̯æː	/	iuu / ïüü ɨ̯ɯː / i̯ɯ̟ː
u	ua / üä ɯ̯ɑ / ɨ̯æ	ui / üï ɯɨ̯ / ɯ̟i̯	/	uaa / üää ɯ̯ɑː / ɨ̯æː	uii / üïï ɯ̯i̠ː / ɨ̯iː	/
aa	/	aai / ääï ɑːɨ̯ / æːi̯	aau / ääü ɑːɯ̯ / æːɨ̯	/	/	/
ii	/	/	iiu / ïïü i̠ːɯ̯ / iːɨ̯	/	/	/
uu	/	uui / üüï ɯːɨ̯ / ɯ̟ːi̯	/	/	/	/

Proto-Rykaic also allows a range of dipththongs. All combinations of two different short vowels are possible, plus all combinations of one short and one long vowel except those with short /a/. In diphthongs where one vowel is long, this is also the prominent vowel. In diphthongs of two short vowels, /a/ is always prominent (hence the lack of diphthongs with short /a/ and a long vowel), otherwise the first vowel has higher prominence. All diphthongs can be front or back, but the two vowels must agree on this feature.

Concluding thoughts

We’ve now created a proto language candidate that should fit with both Ryka and Quriil. On top, it can also serve as a basis for the Konoic languages, possibly placing them in the same language family as Ryka-Quriil.

Since this post has already become much longer than I intended, I have split it into several parts: In the next part, we will have a look at the sound changes we need to actually derive Ryka and Quriil from our new proto language, and how we might have to modify the two to make the relationship plausible. In a third post, I’ll then see how the Konoic languages might fit into this.