Sound Changes Part 2: Ablaut, Metathesis, and Vowel Loss

The last post in this series was about the role that stress-based sound changes play in the development of this language’s triconsonantal roots. This one is about everything else. Or at least, the remaining major¹I say major because I may make some other sound changes when it seems either necessary or interesting. But these are the ones that are going to most radically alter phonology, phonotactics and root structure. sound changes between Old Kyol and High Kyol. 

Ablaut

Ablaut is the simplest of the sound changes I’m discussing today. i pulls preceding vowels /a/ and /o/ to /ə/. This happened before the stress shift, so in some contexts i-ablaut will occur in the descendant language without an i triggering it – contributing even more to root-and-pattern morphology. 

Metathesis

Metathesis is the switching of two consonants within a word to make it easier to pronounce. It’s found most commonly in isolated cases, like english “cavalry” being pronounced “calvary”. But it can occur in grammatical patterns, especially derived stem forms in Semitic languages. In reflexive stems, such as the Hebrew hitpael form and Arabic form VIII, a /t/ in the prefix switches place with the first consonant of the root:

• Hitpael: šdl = hištaddēl (he made an effort)
• Arabic form VIIII: ktb = iktataba (he copied)

(Both examples, sadly, are from Wikipedia – I’m pretty swamped right now and didn’t have time to dig around for a linguistics textbook.)

Metathesis is fairly complicated to evolve. Among linguists there seems to be some debate about how to classify and understand metathesis, the nature of which I am unable to comprehend. That being said, there are some useful trends: Voiceless stops prefer being in onset positions, while continuants (fricatives & liquids) prefer to be in coda positions. Coronal²pronounced with the tip of the tongue touching the mouth, teeth or lips. consonants prefer to go after non-coronal consonants.³ https://www.ling.upenn.edu/~gene/papers/Buckley2011_metathesis.pdf

Let’s apply some of these trends here:

• /t/ switches place with following sibilant fricatives.
• /k/, /g/, /x/ and /ɣ/ switch places with preceding alveolar consonants.
• /ɣ/ and /ʝ/ switch places with preceding stops.

So that we don’t have some extreme weirdness going on, I’m going to place this before the stress-based vowel deletion. This will prevent the most confusing of the effects, where consonants within a triliteral root would regularly switch places -effectively destroying the triliteral root system. However, it does mean that some affixes will end up inside the root:

• ko.’sat.ga → ko.sag.ta → ksag.ta → ksagt

Syllabic Consonants

With the deletion of vowels, we probably already have some syllabic consonants being created. Take an example root r-t-n, proto *rotan. With the -ga suffix, this becomes:

 ra.tan.ga → r.’tan.ga → r.’tang

 It’s probably more reasonable to evaluate the /r/ as its own syllable, rather than trying to cram /rt/ into an onset.

But this isn’t enough syllabic consonants to satisfy my sadistic cravings. If nothing else, it means syllabic consonants will never occur in stressed positions. I simply cannot abide this level of pronounceability!

The process of getting more syllabic consonants is fairly simple. We often consider consonants assimilating into neighbouring vowels, but rarely vowels assimilating into neighbouring consonants. In Mandarin Chinese (and some other Sino-Tibetan languages), some fricatives near a high vowel will assimilate into that vowel, producing “apical vowels” – essentially syllabic voiced fricatives. There are two voiced fricatives, /ʝ/ and /ɣ/, which are already very close to our high vowel /i/. /ʎ/, being a palatal liquid, is also close enough to /i/ to assimilate it. While /ŋ/ at first seems unusual to assimilate an /i/, it’s pronounced as  [ɲ] close to high vowels, and so could also assimilate previous /i/.⁴ Technically, this would create a division between syllabic ŋ (from əŋ) and ɲ (from iŋ). But these two sounds are so close I think that they’ll just recombine into syllabic ŋ If these consonants are in coda position of a syllable after an /i/, they’ll assimilate the /i/ and become syllabic.

In addition, there’s the vowel /ə/. This vowel is very weak. As such, it’ll assimilate into coda liquids. This includes all the consonants which assimilate /i/, as well as /r/, /l/, /m/, and /n/.

In summary:

• i → ∅ / _ ʝ,ɣ,ʎ,ŋ $
• ə → ∅ / _ ʝ,ɣ,r,l,ʎ,m,n,ŋ $

Now something fun happens here. Let’s create the suffix -ɣi, and apply it to our example root *kosag. First, we have umlaut take effect:

• ko.’sat.ɣi → ko.’sət.ɣi

Next, we have metathesis:

• ko.’sət.ɣi → ko.’səɣ.ti

Then, vowel deletion takes place:

• ko.’səɣ.ti → ‘ksəɣt

Finally, the /ɣ/ assimilates the preceding /ə/, giving us

• ‘ksəɣt → ‘ksɣt

Not only does this wonderful word have no vowels whatsoever, but it also just inserted a consonant in the middle of the stem!

Conclusion

But this isn’t the end of the language’s lineage, and the root-and-pattern morphology will develop in its descendant languages. Next time I’ll discuss the sound changes in the daughter language, with a particular eye to the reduction of the consonant clusters created in today’s changes.