Triconstanantal Root Language Families

[Azdusha]

Does anyone have any good sources on tri-constanantal root language families? The concept of such a language is rather cool, but I'm rather confused by how they would be affected by sound changes and how a plausible language family would look. I understand that I can do something like look at hebrew and arabic side-by-side and compare, but I'm really more interested in looking at a mother-daughter connection. Either natlang examples or conlang examples would be nice, as well as advice/tips on how one would make such a conlang family.

[Solarius]

A good thing to look at might be Morrocan Arabic vs. Classical Arabic.

[Astraios]

Aramaic > Neo-Aramaic.

[Yng]

Or Classical Arabic and any dialect, really. There's a very good grammar of Syrian Arabic available online. That said, this won't give you much help if you don't know what you're looking for.

Generally speaking though I'd say that diachronic development affects triconsonantal languages in pretty predictable ways. It helps if you realise that roots aren't really THINGS - I mean, they are THINGS, but they're not necessarily inherent to words and they don't carry any meaning of their own. Rather, you should see derivational processes in triconsonantal languages as operating essentially the same way as these processes in other languages - a source word is plugged in to the formula and a derivation pops out of the other end, deriving its meaning predictably or semi-predictably from the source word, not from the root. The difference in triconsonantal root families - or rather in Semitic, since this is basically a uniquely Semitic-sort-of-Afro-Asiatic feature, although other language families also have discontinuous morphology - is that the consistent factor between the two words is skeletal and largely consonantal in character, with most if not all of the affixes and vowels of the source word being deleted as part of the derivational process.

The question then becomes 'how does the speaker of the language decide what the root is?' and this is an interesting question that often has multiple answers. Whilst it's true that normally affixes are ignored, this isn't necessarily the case - and often, words have multiple possible analyses available. This provides fertile ground for reanalysis. Classical Arabic has innovated some forms like tamaḥwara dawra 'revolve around' < miḥwar 'axis' which has a derivational prefix mi- reanalysed as part of the root, alongside e.g. ʾaššara 'indicate' (derived as if the root were ʾ-š-r from the noun ʾišārah 'indication', even though that is a totally regular verbal noun of the verb ʾašāra, with the apparent root š-y-r) and ʾaslama/yuʾaslimu 'Islamise' derived from ʾislām as though the root were ʾ-s-l-m, again in spite of the fact that ʾislām is a regular verbal noun derived from the verb ʾaslama/yuslimu, apparent root s-l-m.

Viewing the root in this way also allows us to much more easily understand how e.g. the elision of consonants affects the triconsonantal structure - the loss of the glottal stop in all but initial position and the compensatory lengthening of vowels in Syrian Arabic has led to lots of words being reanalysed as having semivowels where Classical Arabic had a stop. Classical Arabic badaʾa yabdaʾu, for example, renders bada yibdi (alongside other variants), which is reanalysed as fitting into the incredibly common conjugation class -a/-i.

On the other side of the coin, analogy and the absorption of other affixes etc can allow the innovation of new ablauts - Syrian Arabic at least has a bunch of reduplicative forms not found in CA, for example, along with forms like faʿlan (I have no idea what the diachronics of this are but it appeared from somewhere). And sound change and analogy can lead to the collapse/disuse of other ablauts - adjective plurals in fuʿāl and fiʿāl have collapsed into one class. Many nouns innovate new plurals, and so on. In some dialects the various different plurals available to adjectives in faʿīl have reduced to fVʿāl/faʿīlīn, for example. Moroccan Arabic innovated a 'verbal diminutive' used for cutesy effect which seems to me to be modelled on nominal diminutives (my vague theory is that verbs were derived on a four-letter pattern from nominal diminutives, which insert a semivowel between the first and second consonants, and that somehow semantic drift and analogy took the resulting verbs and generated a new class).

Hope this helps.

[Drydic]

Does anyone have any good sources on tri-constanantal root language families? The concept of such a language is rather cool, but I'm rather confused by how they would be affected by sound changes and how a plausible language family would look. I understand that I can do something like look at hebrew and arabic side-by-side and compare, but I'm really more interested in looking at a mother-daughter connection. Either natlang examples or conlang examples would be nice, as well as advice/tips on how one would make such a conlang family.

Sound changes are what created triconsonantal languages. They started as something probably quite similar to how we've reconstructed Proto-Indo-European (that is, in how the system functions, not that there's any special closeness between Semitic/Afro-Asiatic and Indo-European), and through vowel reductions, massive analogy, and (this is an important point) retention of the prefixal and suffixal conjugation systems, it became what we'd recognize as Semitic today. And we see the process continue; no Semitic language, even though they all have gone through significant vowel changes over the centuries, has completely lost the root system.
*snips the other bits and pieces she had written since Yng has delivered a masterpiece in the meantime*

[tiramisu]

You may also want to check out Proto-Semitic > Akkadian if you get the chance. Akkadian does a lot of funky things with sound change that can probably answer many of your questions.

For the most part, triconsonantal root languages can be treated like any other language, including in how sound change happens: more or less you apply sound change by morpheme rather than by root. Perhaps a better way to look at these languages -- I think as once suggested by Maknas -- are as languages with very advanced umlaut systems. This way you can see parallels in development with words like sink/sank/sunk, which are derived from the same root but have different vowels. But overall, the unique part of triconsonantal root languages isn't in how they transform -- it's in how they became triconsonantal in the first place.

[kodé]

Hey, don't forget Yokutsan languages, which are slightly less batshit compared to Semitic, but still have loads of cool root-and-template morphology! Plus ejectives, glottalized sonorants, and retroflexes (in some dialects)! Moreover, Yawelmani Yokuts has provided key data for phonological theory for several decades. Bottom line: Yokuts is cool as hell (and way undervalued by conlangers)!

[Nesescosac]

Interesting! Can you recommend cool things to read about it/places I can find out more? Preferably in some easily downloadable format, but other things are nice too.

[Tiamat]

Hey Yng, when are you going to update your thread on triconsonantal languages/non-concatenation morphology?

[Hallow XIII]

critical hit

[Solarius]

Also Aslian languages, which also rely on noncontatenative morphology- and have lots of insane reduplication.

[tiramisu]

Hey, don't forget Yokutsan languages, which are slightly less batshit compared to Semitic, but still have loads of cool root-and-template morphology! Plus ejectives, glottalized sonorants, and retroflexes (in some dialects)! Moreover, Yawelmani Yokuts has provided key data for phonological theory for several decades. Bottom line: Yokuts is cool as hell (and way undervalued by conlangers)!

Tell us moar.