Evolving Triconsonantal Roots

Let’s talk about triconsonantal roots.

In the real world, triconsonantal root systems are very unusual – being exclusive to languages in the Semitic language family (Hebrew, Arabic, etc.) However, they’re also in my conlang. So let’s discuss them and how they arise!

What are triconsonantal roots?

In most languages, words have stems – a string of sounds which carry the word’s lexical meaning. This word’s meaning can be changed by adding prefixes and suffixes to the stem.

Example (English): happy, happier, happiest

Word stem: happy

Affixes: -∅, -er, -est

Occasionally, the sounds within a stem can be altered to convey meaning (a process known as apophony).

drink, drank, drunk

In languages with triconsonantal roots, however, instead of analyzing most words as stems and affixes, we can analyze them as roots and patterns.¹If you know about Semitic languages and are wondering “but what about derived stems?” I’ll get to them another time. The language that I’m working on has a derived stem system, and I’ll go into more detail on that then. Roots are unpronounceable strings – generally of three consonants (hence “triconsonantal”). To inflect them, you put patterns of vowels (and sometimes other consonants) between the roots.

Example (Arabic): kataba, yaktubu

Root: K-T-B

Patterns: 1 a 2 a 3 a, ya 1 2 u 3 u

But this distinction isn’t as split as you might think. In The Origin and Development of Nonconcatenative Morphology²This work is the primary source I’m using here, and one that I’ll frequently use throughout this work. While others have previously come to the conclusion that triconsonantal root systems are simply extended forms of apophony, Simpson’s is one of the most commonly called upon that I’ve seen – probably because it goes into such depth., Andrew Kingsbury Simpson argues that Semitic triconsonantal root systems arise from extremely extended forms of apophony. In other words, the vowel patterns derived from triconsonantal roots have their origins in prefixes and suffixes that have been affected by sound changes. Simpson finds two types of change are at play here – segmentally and prosodically conditioned alternations.

Let’s make up an example verb to see how this works. (I’m not using one from my language, or from a real-world language, as I want some fairly streamlined sound changes to show how this works.)

We’re going to start with a protolanguage that isn’t using a triconsonantal root system – instead, it relies on prefixes and suffixes to inflect its verbs. Let’s say that we have a verb, kadag, and a prefix you can apply to that verb: i-. 

kadag, i-kadag

Now, let’s apply a prosodically conditioned alternation. Vowels in pretonic syllables (syllables directly before the stressed syllable) are ellided unless this would create a word-initial consonant cluster; when this is the case, they are reduced to /ə/.

kədag, i-kdag

(We’re assuming for now that the stress always falls on the last syllable.)

Now, let’s apply a segmentally conditioned sound change. If /i/ pulls non-high vowels in adjacent syllables to /e/, we end up with these forms:

kədag, i-kdeg

This looks a lot like a triconsonantal root system, especially when we compare these forms on other verbs:

*solok → səlok, islek

*noshaz → nəshaz, inshez

Over time, people stop thinking of these as mutations of an original root and instead start analyzing them as different forms of a triconsonantal root. This process is known as analogy.

CəCVC, iCCeC

Words that don’t match this pattern are regularized:

*mot → mot, imet → məmot, immet

Or, if they’re common enough, stick around in their irregular forms:

*kaz → kaz, ikez

You might notice that even in the regular roots this pattern isn’t perfectly consistent. Depending on the second vowel in the root, the first form will have a different second vowel. (Hence CəCVC.) If the second vowel is an /i/, things get even more complicated:

*takir – tekir, itkir

(Note that /i/ won’t affect subsequent /i/, as it’s already as fronted and closed as it can get. Also, note that the prosodically conditioned sound change happens before the segmentally conditioned sound change, so takir → təkir → tekir.)

Because of differences like these, Semitic languages divide verbs into conjugations. These are often based on “thematic vowel”. For this example  language, we’d probably have at least three conjugations based on thematic vowel:

Thematic vowel a: CəCaC, iCCeC

Thematic vowel o: CəCoC, iCCeC

Thematic vowel i: CeCiC, iCCiC

If you want to learn more on this subject, I’d highly recommend this video by Biblaridion. It’s an excellent resource on both triconsonantal roots and nonconcatenative morphology in general.