It’s time to apply some sound changes to our new proto language and see if we can derive Ryka! For Ryka, I already have a lot of vocabulary in place and it’s important that most of it can be derived from the proto language via plausible sound changes. We will first introduce some sound changes to arrive at Proto-Ryka-Quriil, the basis that we will also use in the next part when having a look at Quriil. From this intermediate language, we can then continue towards Old Ryka, which is already very similar to modern (Capital) Ryka.

Before diving into the sound changes from Proto-Rykaic to Proto-Ryka-Quriil, I need to introduce two new phonemes into Proto-Rykaic: The lateral approximants /ʎ/ and /ɫ/. Now we already have them as the vocalic consonants, but while working on the sound changes all the way down to Ryka, I realized that it will make things a lot easier if we also have them as a regular consonant in both onset and coda position. We will see later that they can be used to introduce Ryka’s water glottal and some diphthongs into the phoneme inventory.

I have also decided that Proto-Rykaic had no geminate consonants.

For each sound change (or set of related sound changes), I will introduce a (NAME) that I can use in later discussions to reference them. Proto-Rykaic words will be prefixed with **, Proto-Ryka-Quriil words with *.

Proto-Ryka-Quriil

Sound changes

(VROUND) As the language’s speakers started using the human form more often, some of the vowels became rounded and the central vowels became more centralized.

• ɯ, ɯ̯ → u, u̯
• ɯ̟ → ʉ
• i̠ → ɨ

(CROUND) Alveolar consonants before rounded vowels became labialized:

• tʰ, t, n → pʰ, p, m | _ [+round]

(CFRONT) Most consonants changed their place of articulation depending on whether they came before a “front” /i, ʉ, æ, ʎ̩/ or “back” /ɨ, u, ɑ, ɫ̩/ vowel – note that this means word-final consonants don’t change!

• tʰ, t, n → pʰ, p, m | _ [+front]
• ʈʰ, ʈ, ɳ → tʰ, t, n | _ [+front]
• kʰ, k → qʰ, q | _ [+back]

(VHIGH) The long central vowels merged with their front/back counterparts. The short central vowels merged with each other.

• ɨː → iː
• ʉː → uː
• ɨ, ʉ → ɨ̹

(A) The remaining front-back pair lost its phonemic contrast as well.

• æ, ɑ → a

(CH) Finally, the affricate t͡ɕ became the fricative ɕ.

• t͡ɕ → ɕ

Resulting phoneme inventory

	Labial	Alveolar	Palatal	Retroflex	Velar	Uvular
Aspirated Stop	ph pʰ	th tʰ		ṭh ʈʰ	kh kʰ	qh qʰ
Plain Stop	p p	t t		ṭ ʈ	k k	q q
Fricative			sh ɕ		
Nasal	m m	n n	ñ ɲ	ṇ ɳ	ng ŋ
Approximant			j j		
Lateral approx.			lh ʎ		l ɫ
Trill		r r			

We now arrive at a much higher number of consonants for Proto-Ryka-Quriil: We have five full columns (labial, alveolar, retroflex, velar, uvular) of aspirate stop, plain stop and nasal, plus the alveolar trill and a few palatals.

	Front	Central	Back
High	i / ii i / iː	y ɨ̹	u / uu u / uː
Low		a / aa a / aː	
Vocalic consonant	lh ʎ̩		l ɫ̩

The vowel inventory, on the other hand, has shrunk a bit, as the front-back contrast was (mostly) lost. A remnant of this is the half-rounded central vowel /ɨ̹/. In addition, we have the three vowels /a, i, u/, both short and long, and the vocalic consonants /ʎ̩/ and /ɫ̩/.

	a	i	u	y	aa	ii	uu
a	/	ai1 ai̯	au2 au̯	ay aɨ̯	/	/	/
i	ia1 i̯a	/	iu2 iu̯	iy1 iɨ̯	iaa1 i̯aː	/	iuu1 i̯uː
u	ua3 u̯a	ui1 ui̯	/	uy3 uɨ̯	uaa3 u̯aː	uii3 u̯iː	/
y	ya ɨ̯a	/	/	/	yaa ɨ̯aː	yii1 ɨ̯iː	yuu2 ɨ̯uː
aa	/	aai1 aːi̯	aau2 aːu̯	aay aːɨ̯	/	/	/
ii	/	/	iiu2 iːu̯	iiy1 iːɨ̯	/	/	/
uu	/	uui1 uːi̯	/	uuy3 uːɨ̯	/	/	/

We still have a lot of diphthongs, but the front-back distinction has left strong traces here. Since retracted i /i̠/ and advanced ü /ɯ̟/ have ultimately merged into /ɨ/, the diphthongs in which their advanced and retracted counterparts ï /i/ and u /ɯ/ participate usually occur exclusively in front or back contexts. Combined with the labial split, we get three different restricted patterns in which most of the dipthongs occur. Only those involving Proto-Ryka-Quriil a /a/ and y /ɨ/ may occur after any consonant, since in these cases, a Proto-Rykaic front and back diphthong have merged into one.

Stress in Proto-Ryka-Quriil will go on the first long vowel or diphthong, or on the first vowel when they’re all short. It’s not too different from Ryka in this respect (and, as we will see, Ryka’s stress pattern directly derives from this).

Old Ryka

Now it becomes interesting – what changes do we have to apply to Proto-Ryka-Quriil to arrive at Old Ryka and what will the Proto words look like that we can then reconstruct from the Old Ryka lexicon?

Sound changes

(DENT) Ryka’s coronal consonants are shifted forward in comparsion to its ancestor. To achieve this, we first want to dentalize the alveolars to increase the contrast with the retroflexes:

• tʰ, t, n → t̪ʰ, t̪, n̪

(FRIC) One of the most noticeable differences between Ryka and its Proto language is the large number of fricatives. We can use the aspirated stops to generate these:

• pʰ → ɸ
• t̪ʰ → θ
• ʈʰ → ʂ
• kʰ → x
• qʰ → χ

(MID) To arrive at our five-vowel system, we will gradually lower the non-word-final short high vowels into mid position:

• i → e → ɛ | _ C,V
• u → o → ɔ | _ C,V

(FIN) Word-final nasals in unstressed syllables are just dropped, which will obscure the condition for the vowel lowering sound change a little bit, since we can then also have /ɛ, ɔ/ in word-final position in Old Ryka. Likewise, we drop word-final short vowels without a coda. These sound changes have occurred more or less simultaneously and do not feed into each other, i.e. we cannot have a final nasal deleted from a word and then the now final vowel.

• N → ∅ / V C* V* _ #
• V → ∅ / V C* _ #

(DENAS) Going back to the consonants, the other big difference to Proto-Ryka-Quriil is the absence of nasals (apart from the vocalic consonant), so we will turn the remaining nasals into voiced plosives now that the aspirated-plain contrast is gone and has made room for a voiced-voiceless contrast instead. We don’t have a voiced palatal plosive in Ryka, so we will just turn this nasal into a glide:

• m → b
• n̪ → d̪
• ɳ → ɖ
• ɲ → j
• ŋ → g

(LAT1) Unfortunately, since we lack a uvular nasal in the Proto language, we cannot get the water glottal this way. We will instead use the velarized lateral and move it backwards into the voiced uvular fricative in onsets. Between a consonant and a vowel, on the other hand, both laterals are reduced to semivowels:

• ʎ → j | C _ V
• ɫ → w | C _ V
• ɫ → ʀ | _ V

(LAT2) With the front-back contrast in the laterals mostly gone, all the remaining palatalized/velarized laterals become plain laterals:

• ʎ, ʎ̩ → l, l̩
• ɫ̩ → l̩

(NAS) Now that the language has zero nasals, even though they are a pretty common and easy-to-produce sound, there is a huge gap waiting to be filled. The laterals are ready to jump in, although only in non-onset positions:

• l̩ → n̩
• l → n | _ C

(LEN) Let’s now get rid of the vowel length contrast and apply some compensatory gemination in the process – note that this also serves to explain why Ryka has stress on syllables preceding geminate consonants:

• C → CC / Vː_
• aː, iː, uː → a, i, u

(UVU) The single uvular stop and fricative are weakened into glottal consonants, while their geminate counterparts velarize:

• qː → kː
• q → ʔ
• χː → xː
• χ → h

(ALV) Meanwhile, the retroflex stops gradually become rhoticized/lateralized, merging with the already present alveolar trill /r/ and lateral approximant /l/:

• ʈ → ɽ → ɾ → r
• ɖ → ɭ → l

(SH) The last remaining retroflex consonants, the fricative, merges with the palatal fricative. Note that this is a deviation from the consonant inventory I presented in the previous post for Old Ryka. So far, I’ve always thought that the wind member of the /r/ row evolved from a voiceless /r̥/ into /ʂ/, and then, in modern Capital Ryka, into /ɕ/. The latter had always been a weird sound change, because why would a language without any palatal consonants suddenly palatalize this one fricative? With the palatal fricative already being part of the proto inventory, we can now merge /ʂ/ into it in Old Ryka, providing a good explanation for its presence in Capital Ryka.

• ʂ → ɕ

(CENTR) The central vowel is fronted and merges with /i/. It is lowered to /ɛ/ when preceding the palatal glide or being the prominant part of a falling diphthong for better contrast:

• ɨ̹ → ɛ / _ j, i̯, u̯
• ɨ̹ → i̹

(WATER) Following the consonant mode system, /ʀ/ is inserted before all remaining word-initial vowels:

• ∅ → ʀ / # _ V

Reconstructing some Proto vocabulary

Now let’s get to the real fun: Reconstructing Proto words for the existing Ryka lexicon. To get a feeling for what Proto-Ryka-Quriil and Proto-Rykaic were like and what sound changes I might be missing to fit in most of my Ryka vocabulary, I browsed through my Ryka word list and picked some words that looked interesting, so I might have missed some problematic cases. I did not include functional words (pronouns, articles, etc.) or onomatopoetic vocabulary, since these do not always adhere to the usual sound changes.

So let’s have a look at some of the possible reconstructions:

Meaning	Old Ryka	Proto-Ryka-Quriil	Proto-Rykaic
eye	lu lu	*ṇuu ɳuː	**ṇuu ɳɯː
to support	ped pɛd̪	*pin pin	**tïn tin
tongue	lath laθ	*lhatha ˈʎatʰa	**l̈äṭhä ˈʎætʰæ
mushroom	pot pɔt̪	*put ˈput	**tut ˈtɯt
fire	shyth ɕi̹θ	*shythy ˈɕɨtʰɨ	**chüṭhü ˈt͡ɕɯ̟ʈʰɯ̟
bone	katkhen ˈkaθxɛn	*katkhilh ˈkatkʰiʎ	**käṭkhïl̈ ˈkæʈkʰiʎ
salt	tarku ˈtarku	*ṭaarkuu ˈʈaːrkuː	**ṭäärküü ˈʈæːrkɯ̟ː
feather	‘ypput ʀipːut̪	*liipuut ɫiːpuːt	**liituut ɫi̠ːtɯːt
to open	hyttja ˈhitːi̯a	*qhiitiaa ˈqʰiːti̯aː	**khiitiaa ˈkʰi̠ːtɨ̯ɑː
sick	leuwr lɛu̯r	*ṇiiuṭ ɳiːu̯ʈ	**ṇiiuṭ ɳi̠ːɯ̯ʈ

Impossible Old Ryka words

To test my sound changes more thoroughly and check if there are any Ryka words that cannot be produced through them, I used an old script that I had originally written to reconstruct Proto-Kono words from Asiul or Nyírvon vocabulary. Although the Kono sound changes are deprecated by now, the program works for any set of sound changes written in the syntax of Zompist’s Sound Change Applier. My program compiles them into a Finite State Transducer (FST) and uses the inverse of that FST to generate all possible proto reconstructions for a given modern language word.

Using the sound changes above, my program was able to find proto reconstructions for 304 of 402 tested Ryka words (~75%). Examination of the “impossible” vocabulary provided some very interesting insights regarding the “blind spots” of my sound changes and the phonological patterns that they impose on the modern Ryka language. If I stick with the current set of sound changes, these words will have to be altered, borrowed or considered as irregularities.

High back vowels followed by single consonants

The only way to get Old Ryka /u/ in non-final position is by having a long /ɯː/ (or /ɯ̟ː/) in Proto-Rykaic. However, a long vowel geminates following single consonants on the way to Old Ryka (LEN). Therefore, /u/ followed by a single consonant is impossible in Old Ryka. This renders the following words illegal (among others):

• puka “skin”
• kule “blood”
• gudanth “voyage; day”

The solution for these words is to either geminate the consonant after /u/ (e.g. pukka or kulle) or to lower the /u/ into /ɔ/ so that it can have a short ancestor (e.g. godanth).

Mixed labials/velars and alveolars/glottals

We cannot mix labials/velars and alveolars/glottals in most cases. That is because alveolars and glottals require the Proto-Rykaic word to be back-voweled (CFRONT). Labials and velars almost exclusively appear in front contexts. Hence, a word like bahouwd “story” would have to change to tahouwd or bakhouwd. Some other examples where this combination occurs:

• horkun “knot”
• khaqe “black”
• shojge “to flow”

Some cases of mixed labials and alveolars/glottals actually turned out to be fine, namely when the labial is before a rounded vowel or at the end of the word, since labials can also be derived next to rounded vowels (CROUND):

• poqak “pot” < *puqak < **tukak
• hap “to hunt” < *qhapu < **khatu
• push “mud” < *puṭh/*push < **tuṭh/**tuch
• shyb “glass” < *ṭhimu/*chimu < **ṭhinu/**chinu

/e/ in words with alveolars/glottals

As noted above, alveolars and glottals require back vowels in the Proto-Rykaic ancestor. Short /e/, on the other hand, can only emerge from front contexts: It derives from Proto-Ryka-Quriil short /i/, which in turn derives from Proto-Rykaic short front /i/; the Proto-Rykaic short back /i̠/ is centralized into /ɨ̹/ in Proto-Ryka-Quriil and eventually merges again with /i/ in Ryka, skipping the sound change to /e/. This means that /e/ cannot occur in words with historical back contexts, namely those containing retroflexes and glottals. This affects quite a number of words, for instance:

• beldyl “seed”
• tyhe “help”
• qer “earth”

To fix these words, we either need to eliminate the “back” consonants or replace /e/ by /y/.

/o/ in words with labials/velars

The same mechanism as above prevents the occurrence of /o/ in historical front contexts, as Proto-Rykaic short front /ɯ̟/ also undergoes the change into /ɨ̹/ and then /i/. This mostly affects words with velars, as the combination of /p/ and /o/ can also emerge from back contexts (CROUND). Again, this prevents quite a large number of words, for example:

• kor “tree”
• kerko “to climb”
• pado “rain cloud” (/p/ is not followed by a rounded vowel and thus requires the front context)
• porka “thunder” (while /po/ would also be possible in a back context, /k/ requires front vowels)

To fix these words, we either need to eliminate the “front” consonants or replace /o/ by /y/ (the reflex of Proto-Rykaic /ɯ̟/).

Alveolars followed by /o/

As noted before, labials becomes possible in back contexts when followed by a rounded vowel, as Proto-Rykaic alveolars change into labials before rounded vowels regardless of vowel backness (CROUND). This also means that Ryka alveolars after a rounded vowel should be extremely rare, as all instances of Proto-Rykaic alveolars have labialized in this position. The only way to have a Ryka alveolar followed by a rounded vowel is to have it come from a Proto-Rykaic retroflex in a front context (CFRONT).

Combining this with the restriction above, namely that /o/ can not emerge in front contexts, the combination of a Ryka alveolar and /o/ is obviously impossible. This again eliminates quite a few words, for example:

• thonb “tail”
• tosh “ugly”
• doheyj “big, wide”

These words can be resolved by labializing the offending alveolars or introducing a front-context vowel in place of /o/.

Final mid vowels in monosyllabic words

The sound change that produces /e/ and /o/ is explicitly limited to non-final positions (MID). In words with more than one syllable, we can still get Ryka words ending in /e/ or /o/ when a Proto-Ryka-Quriil word-final nasal was dropped after the vowels changed (FIN). However, since this can only apply to unstressed syllables, monosyllabic words have no chance to get a word-final mid vowel. This affects a few words like:

• bo “spark”
• re “year”
• te “arrow”

These words will either have to get a final consonant or change their vowel to /a/, /i/ or /u/.

Conclusion

Wow, that took quite a long time – almost exactly three months since I defined Proto-Rykaic, even though I already had a rough idea of the sound changes I would need. It was a bit shocking to see that I would have to alter 25% of my Ryka vocabulary to adhere to the new sound changes, and I spend quite some time testing out new complicated sound changes to fix that. In the end though, I discarded most of them because they eliminated other words and seemed implausible and artificial.

I have definitely gained appreciation for the subtle flavors and restrictions sound changes impose on the phonotactics and possible words of a language. When I have fixed the Ryka vocabulary, most of them will be invisible unless someone does some statistical examination of the words, but they will certainly make the language more realistic.

But before making this a fixed thing, let’s first check out Ryka’s siblings and cousins to see how they work out… Starting with the closest relative, the next part of this series will cover the sound changes for Quriil!