I plan to come back and expand on this. It ended up way longer than I anticipated. May contain mistakes at the moment.
I was doing some exploring and was looking through this thread. I'm a year too late, but I have some input. For all I know, this conlang may have been discontinued, but perhaps this will be useful to somebody working on non-concatenative root langs.
First, it needs to be said that I strongly recommend clonelangs for newcomers to non-concatenativity. I think this is helpful for non-newcomers too. But I am a fan of using conlangs to explore new linguistic phenomena (cf. the now archaic term "model language"), so I encourage newcomers especially to risk a venture into shitlangs and to ask for feedback on them, as long as they see it as a learning opportunity and are willing to scrap their conlang altogether. (That is, if Aqmk made a mistake with this thread, it's that he began with an attitude that his conlang was definite, not that he tried and made mistakes.)
My advice for clonelangs:
1) Work from a highly regular and/or synthetic language like Arabic. Don't work from Hebrew. Hebrew is on steroids and will give the average newcomer odd ideas about non-concatenativity. As Winston Churchill once said, "I would let the clever ones learn Arabic as an honor, and Hebrew as a treat." Arabic has many benefits aside from regularity, including its extremely rich lexicon.
2) Keep it simple. Start off with commonly attested linguistic changes, and keep them regular. Getting too complicated will lead to unnatural results. Since vowels are how non-concatenative languages, create regular sound changes for such-and-such vowel in such-and-such environments. Hint: stress, place of articulation of the surrounding consonants, and height/roundedness/frontedness of surrounding nuclei (umlaut) are the typical influences that make a vowel in one word diverge from the same historical vowel in another.
3) If you have trouble not simply recreating an existing Semitlang, and you are uncomfortable with this, look comparatively and historically. Take a look at what other Semitic languages do, and try to figure out the historical development. Lipinski's book on Comparative Semitic Grammar is a great resource for this, because it is meaty, freely available on the internet, and generally well-organized. (Heavily outdated information on Comparative Semitics, but veracity is not important for practicing).
If/when you venture into an a priori non-concatenative lang -- I imagine Aqmk is set on this, at least for this conlang, and I don't discourage it as long as he does not want to flaunt the conlang as realistic -- there should be an entirely different approach.
1)
Root creation. Don't limit yourself to a particular method unless this is an experimental goal of the conlang. The truth is that we still aren't sure where Semitic roots came from. There is a popular theory that triconsonantal roots came from tacking an affix onto a biconsonantal root. I think it caught on because we
know this has happened in many cases. However, the theory doesn't actually hold much weight among Semitists. Although attested in individual words, it's difficult both to demonstrate and conceive en masse.
My suggestion is actually to create a dummy root, sort out your morphology by plugging in that root, and
then go back and create roots. This way you see the big picture, and so can A) determine the most basic morpheme for each root (e.g., in Arabic, this is the 3rd singular masculine stative-perfective) and B) determine the need to create a new root for a certain concept, or if you can derive it using the derivational morphology you've set out.
2)
Word derivation. Don't go crazy here. First, create derivational morphemes. Second, create your derivations based on natural languages and check your derivations against theirs. If you don't, you'll probably end up with very contrived derivations -- or worse: way too many roots. For some reason, a lot of people forget that derivation in non-concatenative languages works just like it does in any other language.
3)
Grammar and Morphology.
- It is key that you start simple, which is the main problem with the conlang at hand. You can make your conlang more complex as you develop it, but you cannot start an a realistic a priori conlang with complex grammar. Syntactic noun cases are fine. Gender, person, and number markers are fine. (I do not recommend overlapping gender/person/number/case. Have a separate morpheme for each.) A simple tense system is fine, i.e. two or three tenses or aspects. Stick with only the indicative mood at first. Basic word order is fine. This should be enough to start forming sentences and getting a feel for the structure of the language.
- Any further work on the grammar should have thought-out considerations for the historical development of the features. This will include reanalysing, reusing, and/or combining existing morphemes in a way that makes sense.
- [**]Semitic took emphatic particle *ʔan and combined it with pronominal suffixes to create personal pronouns. *ʔan + *-ku > *ʔanku 'I'; *ʔan + *-ta > *ʔanta 'you'; etc.
- [**]Arabic reanalyzed the Proto-Semitic ending *-ma (> n), which probably signified the headword of a nominal phrase, to be an indefinite marker.
- [**]Semitic used its causative prefix sa- to mark superlative adjectives as well.
- Altering the vowels of a stem should be done in consideration of regular phonological processes, such as those mentioned at the end of #2 of the clonelang suggestions. Take a look below at how Hebrew ended up with absolute/construct/pronominal 'states,' which Proto-Semitic didn't differentiate in either stem or syntax:
First, here is the list of the vowel changes in Hebrew, which are heavily dependent on stress. Refer to this list as we examine Hebrew nouns.
*a > a / *C_. (unstressed)
*a > i / *C_C.
*a > ɔ / 'C_. (stressed), C_.' (pretonic)
*a > 0 / propretonic, between primary and secondary stress
*i > i / C_C. (unstressed)
*i > e / *'C_ (stressed), C_.' (pretonic)
*i > ɛ / 'CV(C).C_ (post-tonic)
*i > a / *'C_C (stressed)
*i > 0 / propretonic
*i > 0 / #C_.'CV(C)
*i > ɛ / in the environment of a guttural
*u > 0 / C_ (unstressed)
*u > u, o / C_C (I don't remember off the top of my head what environment governs which result, but it's unimportant for us)
*u > o: / 'C_(C) (stressed)
*0 > ɛ / C._C# (=anaptyxis)
*a,i > ɛ / before sonorants
*V > ɛ / *C_C.C#
*0 > i > *C_C.C
'king'
sg absolute: *
'malk-u > malk > 'malɛ
k >
'mɛlɛk
sg pronominal: *
'malk-u-ka >
'malk-
ka
'fruit'
sg absolute: *
'piry-u > 'piry > pi'r
iy >
pri:
sg pronominal: *
'piry-u-ya > 'piry-y > 'piry-
iy >
pir'yi:
'blessing'
sg absolute: *
ba'rak-at-u > bara'kat > brɔ'
kɔ
t >
brɔ'kɔ
h
sg construct: *
ba'rak-at-u + word > bara'kat + word > brka
t + 'word >
birka
t + 'word
tašīm (unhouse) 