Triconsonantal Root Systems

The best topics from Languages & Linguistics, kept on a permanent basis.
hwhatting
Smeric
Smeric
Posts: 2315
Joined: Fri Sep 13, 2002 2:49 am
Location: Bonn, Germany

Post by hwhatting »

Wiseblood wrote:You know, I almost never associate "bi" with "in", but rather "at" or "by/with". I've always thought "fī" better translates "in".
You're right, of course, but I learnt bi "in" when I started to learn Arabic, and the stuff you learn early on unfortunately sticks, even if it's not totally correct... :(

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

Wiseblood wrote:No worries,as I'm quite late getting back to you now. I'm actually just now returning to school after three or four years away. I think I'd actually forgotten what a bump and grind it could be. :-(
Hey, great to see you back! And yeah, I know what you mean...
Wiseblood wrote:How's the Arabic going?
Well, at the moment it isn't. Between school and other things I haven't had a great deal of time.
Wiseblood wrote:Well, I've made a pretty thorough search of the library, as well as done some searching some academic sources, and even spoke with the local comparative Semitic guy - the conclusion I've come to is that there doesn't appear to be any kind of definitive reference work on Proto-semitic. Interestingly enough, a lot of the work being done right now is in Russian (a lot of it seems to be tied to Nostratic, though).
Yeah, I've found comparative analyses tend to have the most useful information. Just yesterday I came across a wonderful comparative grammar (well, wonderful meaning it's the first I've found that's nearly this detailed - "Semitic Languages: Outline of a Comparative Grammar" by Edward Lipiński). I've been looking through it - quite interesting.
Wiseblood wrote:Heh, this is something I'd actually like to research myself someday. The broken plural phenomenon almost certainly dates back to Proto-Semitic
Yeah, I was quite surprised actually when I saw some sources saying that even early Hebrew had broken plurals, though they were later regularized by external endings after vowel loss messed up the system. A good example that still survives is apparently the plural of segolate nouns (eg, Hebrew kelev "dog" ~ klavim "dogs") - apparently many people think that "klavim" is actually derived from a broken plural kilaab.

Anyways, I've definitely come across a number of particularly common patterns and trends in broken plurals (eg, that there are a few patterns that seem to be particularly common in the Semitic languages)
Wiseblood wrote:Hwhatting is right: Arabic suffix pronouns change based on the preceding word's "shape", not on morphological properties. For example, becomes -ya after long vowels: عليّ /ʕalay-ya/ "on me", مستشفاي /mustašfā-ya/ "my hospital", مهندسيي /muhandisī-ya/ "my engineers". Actually, I kind of suspect that these "plural possessives" are really nothing more than the result of similar forms in Hebrew going through the crazy Caananite vowel shift. Do you think you could give me more examples?
Perhaps, since the Hebrew plural possessives are the same as the suffixed forms of some prepositions like `al.

But basically, you've got your singular endings (added to a singular stem):
kalbí "my dog"
kalbechá "your dog"
kalbéch "your dog"
kalbó "his dog"
kalbáh "her dog" (with an historical /h/, not just an orthographic one)
kalbénu "our dog"
kalbechém "your dog"
kalbechén "your dog"
kalbám "their dog"
kalbán "their dog"

In the plural, there's a different set of endings, all added to a plural stem:
klaváy "my dogs"
klavécha "your dogs"
klaváyich "your dogs"
klaváv "his dogs"
klavéhah "her dogs"
klavéynu "our dogs"
klavechém "your dogs"
klavechén "your dogs"
klavehém "their dogs"
klavehén "their dogs"
Wiseblood wrote:You know, I almost never associate "bi" with "in", but rather "at" or "by/with". I've always thought "fī" better translates "in".
Heh, Hebrew's done me over in that regard too! The primary function of be- in Hebrew is "in", with the "by/with" being more of a secondary function.

eritain
Niš
Niš
Posts: 1
Joined: Sun Aug 26, 2007 3:09 pm

Post by eritain »

Wiseblood wrote:
When preceded by ? le- "to", it functions much like an English infinitive: ??? ???? ????? 'ani rotzeh lichtov "I want to write".
Wierd... this somehow strikes me as very "European". Does this construction date back to BH? We tend to avoid these types of constructions in Arabic and use relative clause instead: ???? ?? ???? /'ur?du 'an 'aktaba/ (lit. "I want that I write").
It's been a couple years since my BH classes, but yes, it does. With ? le- "to", the infinitive construct expresses purpose: ????? ???? ?????? "I went down to the river to pray." 1 Samuel 12:23 attests the same form in "cease to pray", and the ? is not characteristic of the verb "cease".

On the other hand, a relative clause construction would likely also be understood just fine.

Oh! From a long time ago: The Ba'al/shame connection. "Shame" is bosheth, which was substituted for Ba'al in names during some anti-idolatrous movement or other when Ba'al was not an approved name for an approved god (probably YHWH). Thus you get Biblical names like Ishbosheth, whose mother probably called him Ish-Ba'al, "man of [the] lord".

P.S.: Erm. Not sure what just happened to all my Hebrew there. I'm new here ... help?

User avatar
Soap
Smeric
Smeric
Posts: 1228
Joined: Sun Feb 16, 2003 2:57 pm
Location: Scattered disc
Contact:

Post by Soap »

Some people with old browsers will have trouble posting Unicode; I really dont know what causes it or how to fix it.
Sunàqʷa the Sea Lamprey says:
Image

User avatar
Nesescosac
Avisaru
Avisaru
Posts: 314
Joined: Tue Jul 31, 2007 10:01 pm
Location: ʃɪkagoʊ, ɪlənoj, ju ɛs eɪ, ə˞θ
Contact:

Post by Nesescosac »

Is "av" one of those aforementioned biconsonantal roots?
I did have a bizarrely similar (to the original poster's) accident about four years ago, in which I slipped over a cookie and somehow twisted my ankle so far that it broke
What kind of cookie?
Aeetlrcreejl > Kicgan Vekei > me /ne.ses.tso.sats/

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

eritain wrote:It's been a couple years since my BH classes, but yes, it does. With ? le- "to", the infinitive construct expresses purpose: ????? ???? ?????? "I went down to the river to pray." 1 Samuel 12:23 attests the same form in "cease to pray", and the ? is not characteristic of the verb "cease".
Ack, sorry for the late reply. Your first example still preserves more or less the original function of the le+infinitive construct; it can be reworded more literally as "I went down to the river for (the purpose of) praying". Your second example, however, looks much more like the modern infinitive.
Aeetlcreejl wrote:Is "av" one of those aforementioned biconsonantal roots?
As in "father"? Then yes, it is. Or you could call it a "primitive root", since the vowel inside of it never changes.

User avatar
Nesescosac
Avisaru
Avisaru
Posts: 314
Joined: Tue Jul 31, 2007 10:01 pm
Location: ʃɪkagoʊ, ɪlənoj, ju ɛs eɪ, ə˞θ
Contact:

Post by Nesescosac »

Are there any words related to "father"?
I did have a bizarrely similar (to the original poster's) accident about four years ago, in which I slipped over a cookie and somehow twisted my ankle so far that it broke
What kind of cookie?
Aeetlrcreejl > Kicgan Vekei > me /ne.ses.tso.sats/

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

Aeetlcreejl wrote:Are there any words related to "father"?
There are, but they're all formed with suffixes:

אבא 'aba "daddy"
אבהות 'abahut "fatherhood"
אבהי 'abahi "paternal"

There's a reason for the additional -h- in there, but I don't remember what it is right off hand.


EDIT: Of course, this is just for Hebrew. Arabic and other Semitic languages may have turned the root into a normal consonantal form.

User avatar
Troelsen
Niš
Niš
Posts: 1
Joined: Fri Nov 16, 2007 11:17 pm
Location: Birmingham, attending school

Post by Troelsen »

This thread is so interesting. I think I'll apply this to my conlang. Yes.
Image
[img]http://i213.photobucket.com/albums/cc126/orangelifestyle/jooo.jpg[/img]

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Grath wrote:Wouldn't sukeret come from "sugar" (sukar, if memory serves me well) rather than from "flood"? That would make more sense to me...
Mecislau wrote:Hmm... On second thought, I think you're right. Again, this is one of the problems of homophonous roots: S-K-R means both "flood" and "sugar". Although actually, both of these would make sense in describing diabetes, just that the u in "sukar" would explain why there's a u in "sukeret".
Maybe the word for "diabetes insipidus" comes from the word for "flood", but the word for "diabetes mellitus" comes from the word for "sugar"?
Mecislau wrote:
Grath wrote:2. What root does mishpacha come from?
Hmm. Good question. The original root was clearly *SH-P-CH, but this doesn't exist as a verb anymore (if it ever did). Some dictionary searching found me shifchah "female slave" from the same root, as well as pach "snare" (possibly related, but I don't know - if it is, it clearly never adopted the she- prefix.
Doesn't the Latin word for "family" come from a word meaning a female servant? (Or, at least, a household servant.)
(stuff about historical linguistics)
What's PSS? I'd have guessed PPS was pre-proto-Semitic, or that PAA was proto-AfroAsiatic, but I can't figure out PSS. "Proto-Southern-Semitic" seems unlikely.
Mecislau wrote:
geoff wrote:A thought about all this: If you start with several biconsonantal roots and a few derivational prefixes and suffixes, aren't you going to end up with several triconsonantal roots which have the first or last consonant in common? Do any relics of anything like this survive in modern Semitic?
Do you mean that, if *sha- may have one been a derivational prefix that merged with the root, there should be many roots still bearing the reflex of this /S/ in the first consonant?
Here's another possibility.
Suppose the protoTRS had prefix *B-, biconsonantal root *C-D, and suffix *-F.
Then the TRS might have a B-C-D root and also a C-D-F root.
Due to drift, their connection wouldn't be transparent; but it might still be there.
Has such a thing happened?
If the proto-TRS also had prefix *G-, biconsonantal root *H-J, and suffix *-K, you could wind up with the following 3Cons roots in the TRS;
B-C-D G-C-D C-D-F C-D-K
B-H-J G-H-J H-J-F H-J-K
with lots of somewhat-fossilized and/or somewhat-drifted and/or somewhat-opaqued semantic connections.
Not only might
lots of roots with C-D- as first and second consonants be connected,
and lots of those with -C-D as second and third consonants might be connected,
but also lots of roots with H-J- as first and second consonants might be connected with lots of those with -H-J as second and third consonants.
Does anything like each of those three situations appear to have happened in Semitic languages?
And, the relation between B-C-D and G-C-D might parallel the relation between B-H-J and G-H-J;
the relation between C-D-F and C-D-K might parallel the relation between H-J-F and H-J-K;
and furthermore, the relation between B-C-D and C-D-F might parallel the relation between B-H-J and H-J-F.
Does anything like each of those three situations appear to have happened in Semitic languages?

-------------------------------------------------------------------------------------
[joke:]
Mecislau wrote:[As I said, reconstructed Proto-Afro-Asiatic looks very IE-esque morphologically. And phonetically. And in many other respects... :?
Well, there's an easy explanation of this. As we're all aware the world is getting bigger. Back when it was small the urheimats of the PIE-speakers and the PAA-speakers were right next to each other. So their languages assimilated to each other.
:mrgreen:
(Yes, I know that's all bullshit, but every piece of it has actually been posted somewhere on the internet with apparent seriousness; I just took a bunch of pre-existing kookiness and combined it.)[/joke]
Last edited by TomHChappell on Thu Jun 04, 2009 10:07 am, edited 1 time in total.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

What about compounds, e.g. compound nouns?

Are there gzarot, mishqalim, or minor binyanim, to handle biconsonantal roots/
What about roots whose middle consonant is the same as one of their other consonants, that is, C1-C1-C2 or C1-C2-C2?
Mecislau wrote:There's only one place in Hebrew where metathesis occurs that I know of, but it's not with the root, but with a root letter and a prefix. It occurs in the past tense of binyan hitpa'el verbs:
In the past tense of hitpa'el, the prefix hit- is added to the root in all of its forms (along with vowel changes and suffixes at the end relaying person/number/gender information).
....
However, if the first letter of the root is one of the so-called 'whistling consonants' ('otiyot shorqot - that is, /s z S tz/), the t of the hit- prefix and the first root letter metathesize (and sometimes a few other languages occur, but that's not important here).
....
If there are prefixes ending in a "whistling consonant" -- for instance /his/-, /hiz/-, /hiS/-, or /hitz/- -- and one of them is used with a root beginning with /t/, does that cause any homophony, polysemy, or ambiguity?
What about a root t-C1-C2 or C1-t-C2, say with a prefix hi-?
Mecislau wrote:Hebrew Intensive Pattern: CiCCeC (Third Person Masculine Singular Past Tense Intensive)
Akkadian Intensive Pattern: uCaCCaC (First Person Singular Future Tense Intensive).
As you can see, the intensive is usually formed by geminating the second consonant.
hwhatting wrote:I don't know whether I get you right here...
As far as I can see, there are no consonant clusters in roots, at least in Arabic, i.e. there are no roots that would be *kt-b-m (there's no /v/ in Standard Arabic). If you have a word X-kt-Y, with X and Y being consonants, this is either from a root x-k-t plus suffix Y or from a root k-t-Y with prefix X. It would normally not be from a root x-kt-Y (there are 4-consonant roots, but theyr're very rare).
Mecislau wrote:Well, in Hebrew at least, four-consonant roots are definitely not rare. Five- consonant roots are, though they do occasionally appear. However, you never see clusters in roots with three or less consonants. Biconsonantal roots are always C-C, and triconsonantal always C-C-C. Quadraconsonantal roots always treat the middle two as one (ie, C-CC-C). 5-cons always treat the first two as one (CC-CC-C).
How do you form the intensive of a tetra-consonantal root? For instance, suppose /b-df-g/ is a root, do you get /b-dfdf-g/ in the intensive?
How do you form the intensive of a biconsonantal root?
If there are pre-affricated or post-affricated or pre-aspirated or post-aspirated or pre-nasalized or post-nasalized consonant phonemes in an Afro-Asiatic Tri-Consonantal-Root-System language, how do speakers and hearers distinguish the "complex consonant phonemes" from a consonant-cluster?

hwhatting
Smeric
Smeric
Posts: 2315
Joined: Fri Sep 13, 2002 2:49 am
Location: Bonn, Germany

Post by hwhatting »

TomHChappell wrote: Doesn't the Latin word for "family" come from a word meaning a female servant? (Or, at least, a household servant.)
Yes, familia is related to famulus "servant, slave". But it seems that the original meaning of famulus was "member of the household" and that it is derived from Oscan faama "house". See also French domestique from Latin domus "house" for a similar development "belonging to the house(hold) -> "servant".

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

TomHChappell wrote:Maybe the word for "diabetes insipidus" comes from the word for "flood", but the word for "diabetes mellitus" comes from the word for "sugar"?
Nah, it's the same word, and it's pretty transparently derived from "sugar". Diabetes insipidus is apparently known as סוכרת תפלה sukeret tfelah, literally "bland diabetes" (a calque of insipidus, I guess).
TomHChappell wrote:What's PSS? I'd have guessed PPS was pre-proto-Semitic, or that PAA was proto-AfroAsiatic, but I can't figure out PSS. "Proto-Southern-Semitic" seems unlikely.
"Pre-Proto-Semitic" + typo.
TomHChappell wrote:What about compounds, e.g. compound nouns?
Well, in Hebrew at least, those are generally handled using the construct state of one noun followed by the absolute or definite state of another. In some cases these are hyphenated, in others not. These are pluralized by pluralizing the construct noun, or made definite by converting the absolute state noun to the definite state.

בן-אדם ben-'adam "human being" (lit. "son of Adam"): son.CONST.SG-Adam.ABSOL.SG
בני-אדם bney-'adam "human beings" (lit. "sons of Adam"): con.CONST.PL-Adam.ABSOL.SG

תחנת רכבת tachanat rakevet "train station": station.CONST.SG train.ABS.SG
תחנת הרכבת tachanat harakevet "the train station": station.CONST.SG train.DEF.SG

There are a few more interesting methods; however, they are much less frequent. Occasionally you do come across true compounds that are just two nouns tacked together and treated as one, as in קולנוע qolnoa` "cinema" (from קול qol "voice" + נוע noa` "movement"). You can also get portmanteaux, such as רמזור ramzor "traffic light", from the verb רמז ramaz "to indicate" + אור 'or "light", with the glottal stop elided.

Hebrew in particular is very fond of initialisms too, such as סכו"ם sakum "cutlery, silverware", from סכין כף ומזלג sakin, kaf umazleg "knife, spoon, and fork".
TomHChappell wrote:Are there gzarot, mishqalim, or minor binyanim, to handle biconsonantal roots/
Yes. In traditional Semitic grammatical notation biconsonantal roots are classified as having C2 be /w/ or /j/, so roots such as קם qam "to get up" is Q-W-M, and שר shar "sing" is SH-Y-R, though in practice this is really just a matter of convenience in order to keep with the general three-consonant pattern. More modern/technical descriptions will describe these roots as *qūm and *šīr, respectively, with an inherent vowel.
TomHChappell wrote:What about roots whose middle consonant is the same as one of their other consonants, that is, C1-C1-C2 or C1-C2-C2?
That is quite common, yes, and they have their own conjugation pattern. Historically many of these types of roots (C2=C3) come from Proto-Semitic biconsonantal roots where the second consonant was geminated.

Semitic languages generally show a lot of alternation between biconsonantal roots and roots with C2=C3, as it appears these were more or less interchangeable (eg, Proto-Semitic *subb <--> *sūb "turn", with the length freely being transferred between the consonant and vowel). As a result, many roots are historically attested both ways (eg, Biblical Hebrew biconsonantal סב sav and triconsonantal סבב savav).

Hebrew at the very least has incorporated this productively for some roots, which are biconsonantal in binyan pa`al but triconsonantal in others (eg, חם cham "[be] hot" > חימם chimem "heat, warm up". I believe other Semitic languages have done so as well to varying degrees.
TomHChappell wrote:If there are prefixes ending in a "whistling consonant" -- for instance /his/-, /hiz/-, /hiS/-, or /hitz/- -- and one of them is used with a root beginning with /t/, does that cause any homophony, polysemy, or ambiguity?
In theory, I suppose, but this doesn't really come up. The only one of those that is real is /hiS/-, which only appears on a very small set of Aramaic loan verbs.
TomHChappell wrote:What about a root t-C1-C2 or C1-t-C2, say with a prefix hi-?
That's not really a problem at all. This happens quite often (eg, T-CH-L > התחיל hitchil "begin, start"), but it's clear that the prefix here is hi-, not hit-, because if you remove the hit- you only have two consonants left. Hif`il and Hitpa`el do not tolerate biconsonantal roots; they have to be augmented into triconsonantal roots somehow, such as by reduplicating the final consonant as I described above.
TomHChappell wrote:How do you form the intensive of a tetra-consonantal root? For instance, suppose /b-df-g/ is a root, do you get /b-dfdf-g/ in the intensive?
In Hebrew, and I believe for the other Semitic languages as well, tetraconsonantal roots are inherently considered intensive. It's a perfect "trick": the middle consonant of a triconsonantal intensive is doubled, so for a tetraconsonantal root you just stick in two consonants in that slot. That is, where a triconsonantal root has C1iC2C2eC3 (for Hebrew pi`el), a tetraconsonantal root has C1iC2C3eC4.
TomHChappell wrote:How do you form the intensive of a biconsonantal root?
Via augmentation: the biconsonantal root is converted into a triconsonantal root. This can be done in one of two ways:

1) Reduplicating the final consonant, so, for example, CH-M "[be] hot" becomes CH-M-M "heat".

2) Converting the inherent vowel that comes with all biconsonantal roots into a consonant. Hebrew has only two such vowels: /i:/ > /j/, /u:/ > /v/ (originally /w/). Other Semitic languages still preserve /a:/ as a possible root vowel, which generally becomes /?/ or /h/. I can't think of a Hebrew verbal example of this off the top of my head, but there is מת met "die" (Proto-Semitic *mūt) > מות mavet "death". If I remember correctly, this process is very productive in other Semitic languages such as Arabic, but in Hebrew has largely lost its productive force; however, this type of augmentation did convert many formerly biconsonantal roots into triconsonantal ones in Hebrew, such as Proto-Semitic *šāl "ask" > Hebrew שאל sha'al "ask" (root *SH-'-L).
TomHChappell wrote:If there are pre-affricated or post-affricated or pre-aspirated or post-aspirated or pre-nasalized or post-nasalized consonant phonemes in an Afro-Asiatic Tri-Consonantal-Root-System language, how do speakers and hearers distinguish the "complex consonant phonemes" from a consonant-cluster?
Well, first of all, the triconsonantal system is only found in Semitic, not all of Afro-Asiatic, and Semitic has much less of that. In fact, the closest thing I can think of is how Modern Hebrew treats the affricate /ts/ as a single consonant. Some of those stranger features may be found in the South Arabian and African Semitic languages, however. While I imagine confusion might come up every now and then, I can't really see it being a problem. After all, what language doesn't have its ambiguity?

I can think of a few cases of ambiguity in Hebrew resulting from consonants dropping or merging, but that doesn't seriously harm the language's triconsonantal system, it just results in some reanalysis of the structures of individual roots. For example, Hebrew's (not just Modern, but historical) habit of dropping word-final glottal stops has resulted in some confusion between roots with final {?} and final {h}. For example, the root *Q-R-' "read, call, invite" has two passive participles in pa`al: the expected קרוא qaru' "invited", and the unexpected קרוי qaruy "named" that would be expected from a hypothetical root *Q-R-H. On the other hand, the real root *Q-R-H "happen, occur" has two different forms of the feminine singular of its active participle existing in free variation, the expected קורה qorah and the unexpected קורית qoret which was acquired from the final *' paradigm.

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Thanks very much, Mecislau; you've answered all the questions in my second post and the top half of my first post. (And might I add, answered them completely and clearly? Or should I just say "comprehensibly and comprehensively"?)
Mecislau wrote:Well, first of all, the triconsonantal system is only found in Semitic, not all of Afro-Asiatic,
I had been under the impression that although all Semitic languages were TRS and all living (and nearly all known dead) TRS languages were Afro-Asiatic, there were AA languages that weren't TRS and TRS languages that weren't Semitic. For instance see this.
Wikipedia says "templatic morphology" is a "common feature" of AA languages.
They wrote:a templatic morphology in which words inflect by internal changes as well as with prefixes and suffixes
And they give examples in Arabic(Semitic), Berber(Berber), Somali(Cushitic), Beja, and Hausa(Chadic).
Wikipedia doesn't count Egyptian as Semitic, yet it wrote:Egyptian is a fairly typical Afro-Asiatic language. At the heart of Egyptian vocabulary is a root of three consonants. Sometimes there were only two, for example <rʕ> /riʕa/ "sun" (where the [ʕ] represents a voiced pharyngeal fricative), but larger roots are also common some being as large as five /sḫdḫd/ "be upside-down". Vowels and other consonants were then inserted into the consonantal skeleton in order to derive different meanings, in the same way as Arabic, Hebrew, and other Afro-Asiatic languages do today.
Wikipedia counts Tamasheq as a Tuareg(Berber) language, not a Semitic language. Yet they wrote:As a root-and-pattern, or templatic language, triliteral roots (three-consonant bases) are the most common in Tamasheq. Niels and Regula Christiansen use the root k-t-b (to write) to demonstrate past completed aspect conjugation:
I looked into the subfamilies of the subfamilies of A-A; I don't have time to look further.
I know "templatic morphology" doesn't imply "triconsonantal root system"; in fact it doesn't even imply that the "transfix", "simulfix", or "runaway apophony" or "runaway ablaut" part of the morphology dominates the prefix, suffix, and infix part. Nevertheless there do appear to be non-Semitic languages in the A-A family that are indeed "triconsonantal root system" languages.

Unless all those guys I just quoted are wrong? I have no way of telling how reliable they are.
Mecislau wrote:and Semitic has much less of that.
Quite possibly (i.e. FAIK) the same is true of all the TRSs in A-A, even the non-Semitic ones.
Some A-A languages have quite a few such consonant phonemes with complex articulation; but ANAICT none are TRSs.
Can anyone who's posted to this thread find out? Thanks.
Mecislau wrote:In fact, the closest thing I can think of is how Modern Hebrew treats the affricate /ts/ as a single consonant. Some of those stranger features may be found in the South Arabian and African Semitic languages, however. While I imagine confusion might come up every now and then, I can't really see it being a problem. After all, what language doesn't have its ambiguity?
Thanks!

-------------------------------------------------------------------------------------
These guys wrote:Phonological requirements on regularity are not restricted to Tamil. Arabic and Hebrew have disyllabic requirements for verbs. In Arabic, not only do the verbs have to be disyllabic and triconsonantal, but there are also constraints on what consonants make up the triliteral root. Obstruents and liquids are acceptable, but glides are not. If a glide is one of the three consonants that make up the root, then the verb stem is irregular.
OK, now that you've answered those questions, I have more.
* Do A-A TRSs (especially Semitic languages) commonly have consonants that never occur in roots, but only in prefixes or suffixes?
* If so, are there any commonalities or general tendencies among the languages, as to which consonants or what types of consonants those are?
* Do A-A TRSs (especially Semitic languages) commonly have consonants that never occur in prefixes or suffixes, but only in roots?
* If so, are there any commonalities or general tendencies among the languages, as to which consonants or what types of consonants those are?
* Are prefixes and suffixes likelier to contain just one consonant than more than one? (In most A-A TRSs, or in most Semitic languages.)
* If not, are prefixes and suffixes likelier to contain at most two consonants than more than two? (In most A-A TRSs, or in most Semitic languages.)
* For either of those last two questions, is the answer the same for prefixes as it is for suffixes? (In most A-A TRSs, or in most Semitic languages.)
* In these languages, are vowels earlier than the first consonant of the root, ordinarily consider part of the prefix instead of part of the root?
* In these languages, are vowels later than the last consonant of the root, ordinarily consider part of the suffix instead of part of the root?
* Are there, in these languages, commonly consonants that can appear elsewhere in the root, but are dispreferred as the first consonant of the root?
* Are there, in these languages, commonly consonants that can appear elsewhere in the root, but are dispreferred as the last consonant of the root?
* Are there, in these languages, commonly consonants that can appear elsewhere in the root, but are dispreferred as the middle consonant of the root?
* If the answer to any of those last three questions is "yes", are there any commonalities or general tendencies through these languages?
* Are there, in these languages, commonly consonants that can appear elsewhere in a prefix, but are dispreferred as the last consonant of a prefix?
* Are there, in these languages, commonly consonants that can appear elsewhere in a suffix, but are dispreferred as the first consonant of a suffix?
* If the answer to either of those last two questions is "yes", are there any commonalities or general tendencies through these languages?
* Is there commonly a limit on the length of vowel-clusters?
* If so, what is that limit likely to be?
* Which morphological processes are likely to lead to long vowel-clusters?
* What, if any, common types of constraints are likely to be in effect concerning two consonants appearing in a root?
** That is, for example, is it dispreferred to have all three consonants in the root be voiced?
** Or dispreferred for all three to be unvoiced?
** Is it dispreferred to have two of the consonants be the same, either C1C1C2 or C1C2C1 or C2C1C1?
** Is it dispreferred for two of the consonants of the root to be at the same PoA?
** Is it dispreferred for two of the consonants of the root to have the same MoA?
* How about asking the same questions about prefixes instead of roots, or about suffixes instead of roots?
* How common are infixes in these languages?
** What are some, if some of them have some?
** Do infixes usually follow the first consonant or precede the last consonant?
** Do infixes usually consist of just one consonant?

-------------------------------------------------------------------------------------
Obviously I expect it might take more time and effort than you have readily available to answer all of those questions with the same high-quality of answer you've been pleased to provide us so far.
So let me say that any answer at all will be appreciated; and it will be appreciated at any time.

---------
Thanks
-----
Tom

-------------------------------------------------------------------------------------
And BTW, Hans,
hwhatting wrote:Yes, familia is related to famulus "servant, slave". But it seems that the original meaning of famulus was "member of the household" and that it is derived from Oscan faama "house". See also French domestique from Latin domus "house" for a similar development "belonging to the house(hold) -> "servant".
Thanks.

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

I may be wrong about other Afro-Asiatic languages, but generally Semitic is considered to be the closest to having a system where a root can be analyzed as only three consonants and where conjugation is performed by modifying the vowels between them; it isn't technically quite at this point, but it's fairly close.

Other Afro-Asiatic languages, as I understand, may have roots consisting of three consonants, but not three completely unlinked consonants. That is, Semitic has *K-T-B, while Berber (for example) has ktVb - three consonants, but not all unlinked from one another. The number of vowel combinations is also far more limited.
TomHChappell wrote:
These guys wrote:Phonological requirements on regularity are not restricted to Tamil. Arabic and Hebrew have disyllabic requirements for verbs. In Arabic, not only do the verbs have to be disyllabic and triconsonantal, but there are also constraints on what consonants make up the triliteral root. Obstruents and liquids are acceptable, but glides are not. If a glide is one of the three consonants that make up the root, then the verb stem is irregular.
That's... not entirely accurate. Sure, Arabic verbs may be disyllabic in MSA, but that's just because MSA always has non-zero verbal suffixes (IIRC). Modern Arabic dialects, though, frequently are only a single syllable. Glides are also acceptable parts of roots, but due to historical sound changes they tend to be very "weak" and prone to dropping, unless they were geminated. Most of the "irregular" paradigms in modern Semitic languages were at one time regular.
TomHChappell wrote:OK, now that you've answered those questions, I have more.
Holy crap. I have no problem going through all these (some are quite interesting questions), but don't expect all of them at once!
TomHChappell wrote:* Do A-A TRSs (especially Semitic languages) commonly have consonants that never occur in roots, but only in prefixes or suffixes?
* If so, are there any commonalities or general tendencies among the languages, as to which consonants or what types of consonants those are?
* Do A-A TRSs (especially Semitic languages) commonly have consonants that never occur in prefixes or suffixes, but only in roots?
* If so, are there any commonalities or general tendencies among the languages, as to which consonants or what types of consonants those are?
In all the Semitic languages I know much about, no, there are no restrictions. And aren't restrictions like these extremely rare worldwide anyways? Generally phonemes aren't limited to certain grammatical categories or forms.

Hebrew does have one weird quirk, but I'm not completely sure if it's historical or a "side-effect" of its modern revival. Generally, the language is completely intolerant of word-final consonant clusters in native words, but the 2sg fem form of the past tense is formed with the ending -t, which usually ends up forming a cluster. This results from the loss of a former word-final heavily reduced vowel. However... this isn't really relevant to your question anyways.

The Semitic languages, though, do have a very small set of "afformatives" (consonantal affixes). In fact, for verbs I can list them all: PS *mV- (nominalizing), *nV- (detransitive), *tV- (reflexive), *šV- (causative), not counting personal agreement affixes. The great variety of different forms seen in the modern Semitic languages were the result of combining these with different vowel templates inherited from Afro-Asiatic, and by combining multiple afformatives together to combine their meanings. Take the Arabic conjunction staf`ala, which includes both the causative and the reflexive afformatives. In fact, different Semitic languages ended up adopting different combinations of afformatives, leading to the variety of verb conjugations seen from one language to another.

(The frequent hi- in Hebrew verbs or 'i in Arabic ones are epenthetic, added to block the initial cluster formed by the afformative. For example, PS *t-pa`ala > Hebrew hitpa`el)
TomHChappell wrote:* Are prefixes and suffixes likelier to contain just one consonant than more than one? (In most A-A TRSs, or in most Semitic languages.)
* If not, are prefixes and suffixes likelier to contain at most two consonants than more than two? (In most A-A TRSs, or in most Semitic languages.)
Depends. All the Semitic afformatives and most of the personal affixes are a single consonant. A few of the personal affixes have multiple consonants, but that just reflects their pronominal origins, such as Hebrew 'atem katavtem "you (MPl) wrote". However, as I described above, the Semitic languages frequently combined multiple afformatives together.

I know that this is not true for all AA languages, though. I know that Tamazight (Berber), at least, has a number of affixes consisting of multiple consonants, but I know very little about their history. All I remember is the 3pl ending for the particular dialect I was reading about consisted of something like a five-consonant cluster.
TomHChappell wrote:* For either of those last two questions, is the answer the same for prefixes as it is for suffixes? (In most A-A TRSs, or in most Semitic languages.)
It varies from language to language. Hebrew and Arabic generally don't have more than two consonants in either prefixes or suffixes. However, Modern Aramaic (IIRC) has some longer prefixes and suffixes resulting developing new verbal forms from pre-verbs (reduced forms of other verbs) + personal affixes + participles (which often have the *mV- afformative).



I'll come back to the other questions later.

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

Oh, I forgot to mention that there are, however, dissimilatory processes in the vowels in prefixes. For example, Barth's Law (which applies to a number of west Semitic languages) states that the vowel in the personal prefixes of the imperfective should be as far away from the vowel inside the stem (always of the form -CCVC- in the imperfective) as possible. That is, if the stem vowel is /i:/, the prefix vowel should be /a/; if the stem vowel is /a:/, the prefix vowel should be /i/. You can still see this in Hebrew:

יכתוב yikhtov "he will write" (-o- < PS *-ū-, I think)
יכתיב yakhtiv "he will dictate"

(The vowel in the imperfective was originally inherent to each root. Hebrew, however, opted for maximum distinction by generalizing the vowels to various forms. Pa`al, for example, usually gets -o- < *-ū-, and rarely -a- < *-ā-. Hif`il always gets -i- < *-ī-, etc)

User avatar
Salmoneus
Sanno
Sanno
Posts: 3197
Joined: Thu Jan 15, 2004 5:00 pm
Location: One of the dark places of the world

Post by Salmoneus »

Mecislau wrote:
TomHChappell wrote:
TomHChappell wrote:* Do A-A TRSs (especially Semitic languages) commonly have consonants that never occur in roots, but only in prefixes or suffixes?
* If so, are there any commonalities or general tendencies among the languages, as to which consonants or what types of consonants those are?
* Do A-A TRSs (especially Semitic languages) commonly have consonants that never occur in prefixes or suffixes, but only in roots?
* If so, are there any commonalities or general tendencies among the languages, as to which consonants or what types of consonants those are?
In all the Semitic languages I know much about, no, there are no restrictions. And aren't restrictions like these extremely rare worldwide anyways? Generally phonemes aren't limited to certain grammatical categories or forms.
But it isn't rare at all to have a more expansive array of phonemes available for the "main" (usually first, iirc) syllable, nor for the "main" syllable to have to be in the root. (ie sounds occur in the root that cannot occur in affixe because the affixes can never be the main syllable).
Blog: [url]http://vacuouswastrel.wordpress.com/[/url]

But the river tripped on her by and by, lapping
as though her heart was brook: Why, why, why! Weh, O weh
I'se so silly to be flowing but I no canna stay!

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Mecislau wrote:---(all the other things you wrote)---
Thanks very much! These are informative and pertinent answers.
Mecislau wrote:Holy crap. I have no problem going through all these (some are quite interesting questions), but don't expect all of them at once!
....
I'll come back to the other questions later.
Obviously I expect it might take more time and effort than you have readily available to answer all of those questions with the same high-quality of answer you've been pleased to provide us so far.
So let me say that any answer at all will be appreciated; and it will be appreciated at any time.
Mecislau wrote:Hebrew in particular is very fond of initialisms too, such as סכו"ם sakum "cutlery, silverware", from סכין כף ומזלג sakin, kaf umazleg "knife, spoon, and fork".
By "initialisms" you mean "acronyms", right? (I know "acronym" is, itself, a neologism. What's Hebrew for "acronym"?)

-------------------------------------------------------------------------------------

New question:
About compounds, Mecislau wrote:Well, in Hebrew at least, those are generally handled using the construct state of one noun followed by the absolute or definite state of another. In some cases these are hyphenated, in others not. These are pluralized by pluralizing the construct noun, or made definite by converting the absolute state noun to the definite state.
....
There are a few more interesting methods; however, they are much less frequent. Occasionally you do come across true compounds that are just two nouns tacked together and treated as one, ... . You can also get portmanteaux, ... .
Do any tetraconsonantal roots come from compounds of two biconsonantal roots? Do any quinquiconsonantal roots come from compounds of a biconsonantal root with a triconsonantal root, in either order?

-------------------------------------------------------------------------------------

Thanks.

-------------------------------------------------------------------------------------
Mecislau wrote:And aren't restrictions like these extremely rare worldwide anyways? Generally phonemes aren't limited to certain grammatical categories or forms.
Well, to start with, most if not all of what I've read about these questions is more about statistics and probabilities than about absolute exclusions; I guess I should have phrased my question that way. I've read several things, mostly kind of similar to what Salmoneus said (and I apologize for not having references for these). Here are what I can remember, IIRC.

* I read that sounds that are preferred in the language as a whole tend to be even more strongly preferred in the roots of lexical-content words and somewhat less strongly preferred in affixes and structure words (function words) and particles; and that sounds that are dispreferred in the language as a whole tend to be even more strongly dispreferred in the roots of lexical-content words and somewhat less strongly dispreferred in affixes and structure words etc.

* I have read that (in languages with many two-or-more-syllable words) the statistics governing first syllables of words are often different from those governing subsequent syllables, and the statistics governing last syllables are often different than from those governing prior syllables (that's the same fact unless there are some three-or-more-syllable words).
(That allows the speaker to make educated bets about where word boundaries are likely to be. Of course, so does stress-distribution, in most languages. This is a little different from what Sal said; he was talking about the syllable with primary stress, whereas the thing I read was talking about the first syllable, or about the last syllable; but as Sal points out in many languages the primary-stress syllable will be either the first or the last syllable of the word or either the first or last syllable of the root.)
And (if there are enough three-or-more-syllable words), I've seen it proposed that the statistics governing the first two syllables are different from those governing subsequent syllables, and/or that the statistics governing the last two syllables are different from those governing prior syllables. (These statements will all be mutually non-equivalent only if there are significantly many five-syllable-or-longer words.)

* I have read that, in words-as-a-whole, phonotactics is mostly governed by which pairs of sounds can be consecutive; but in roots, there are often high-ish priority constraints about whether or not certain pairs of sounds can both be in a root, even if some other sound comes between them. One of the examples mentioned was, explicitly, Semitic roots.
I've also read something specifically about Arabic roots, that some consonants strongly preferred for first root-consonant were strongly dispreferred for middle root-consonant and/or vice-versa; and/or that some consonants strongly preferred for first root-consonant were strongly dispreferred for last root-consonant and/or vice-versa.

- - - - - - - - - - - - -

If you (or anyone) knows that any of these are true, or know that any are false, I'd like to read you say so. And if anyone (including you) knows of any references on these questions I'd appreciate reading about them.

For the questions about Arabic rather than Hebrew, perhaps Wiseblood or one of the other major contributors to this thread will chime in?

User avatar
Salmoneus
Sanno
Sanno
Posts: 3197
Joined: Thu Jan 15, 2004 5:00 pm
Location: One of the dark places of the world

Post by Salmoneus »

Major/minor syllable dichotomies are common in southeast asia. In the Mon-Khmer languages, the second syllable is the major one. For example, in Cambodian the first syllable is a consonant followed by a schwa; the second syllable can be C(C)(C)V(V)C.

Iirc, many Tai-Kadai languages, on the other hand, allow complex initial-syllable clusters in an otherwise cluster-free phonology, and I believe that at least some of them allow 'minor' prefix syllables as well.

[The Austro-Tai Hypothesis says that these clusters derive from the reduction of unstressed PAT initial syllables]
Blog: [url]http://vacuouswastrel.wordpress.com/[/url]

But the river tripped on her by and by, lapping
as though her heart was brook: Why, why, why! Weh, O weh
I'se so silly to be flowing but I no canna stay!

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Salmoneus wrote:Major/minor syllable dichotomies are common in southeast asia. In the Mon-Khmer languages, the second syllable is the major one. For example, in Cambodian the first syllable is a consonant followed by a schwa; the second syllable can be C(C)(C)V(V)C.
Thanks, Sal.
Salmoneus wrote:Iirc, many Tai-Kadai languages, on the other hand, allow complex initial-syllable clusters in an otherwise cluster-free phonology
Don't some Balto-Slavic languages allow onset clusters only in first syllables and coda clusters only in last syllables? Not allowing any other intrasyllabic clusters?

- - - - - - - - - - - - - - - - - - - -

Ob-TRS: Do such things, or similar things, also occur in one or more TRS languages?
Mecislau wrote:dissimilatory processes in the vowels
I'd like to read more about these. Where can I do so?
Mecislau wrote:Other Afro-Asiatic languages, as I understand, may have roots consisting of three consonants, but not three completely unlinked consonants. That is, Semitic has *K-T-B, while Berber (for example) has ktVb - three consonants, but not all unlinked from one another. The number of vowel combinations is also far more limited.
Hmm. I see what you mean.
I guess I don't know, other than just taking your word for it; and at the moment I'm not sure how I could check.
Thanks.
Mecislau wrote:
These guys wrote:Phonological requirements on regularity are not restricted to Tamil. Arabic and Hebrew have disyllabic requirements for verbs. In Arabic, not only do the verbs have to be disyllabic and triconsonantal, but there are also constraints on what consonants make up the triliteral root. Obstruents and liquids are acceptable, but glides are not. If a glide is one of the three consonants that make up the root, then the verb stem is irregular.
That's... not entirely accurate. Sure, Arabic verbs may be disyllabic in MSA, but that's just because MSA always has non-zero verbal suffixes (IIRC). Modern Arabic dialects, though, frequently are only a single syllable. Glides are also acceptable parts of roots, but due to historical sound changes they tend to be very "weak" and prone to dropping, unless they were geminated. Most of the "irregular" paradigms in modern Semitic languages were at one time regular.
It's not clear from your reply that you understood that abstract was about what was required for a verb-stem to be regular. It wasn't saying there aren't exceptions to those constraints; it was saying that if those constraints are violated the verb-stem is irregular.
Maybe you understood that already; from your reply I can't be sure either way.

---------

Once again:
Thanks all!

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

Sorry for the delay.
* In these languages, are vowels earlier than the first consonant of the root, ordinarily consider part of the prefix instead of part of the root?
* In these languages, are vowels later than the last consonant of the root, ordinarily consider part of the suffix instead of part of the root?
By "root" I assume you mean "stem", right?

(The root is the consonantal base, such as "K-T-B", that is an abstract concept with no surface realization. The stem include vowels that have been filled in between the root consonants, to which affixes are added, such as the Hebrew past tense stem *katav- and future tense stem *-khtov-)

In that case, no, such vowels are not considered part of the stem.
* Are there, in these languages, commonly consonants that can appear elsewhere in the root, but are dispreferred as the first consonant of the root?
* Are there, in these languages, commonly consonants that can appear elsewhere in the root, but are dispreferred as the middle consonant of the root?
* Are there, in these languages, commonly consonants that can appear elsewhere in the root, but are dispreferred as the last consonant of the root?
* If the answer to any of those last three questions is "yes", are there any commonalities or general tendencies through these languages?
Hmm. That's an interesting question. I can pretty much safely say that yes, there are definitely some preferences for what consonants can occur where, but I cannot really give you any details. I've never really looked into this that much.

However, the majority of any such trends are most likely just historical accident. Medial consonantal {w} and {j} are fairly rare, for example, due to the consonants' general weakness and their early merger with the biconsonantal paradigms. Final {h} not resulting from an earlier *w or *j is also very rare if not nonexistent.

Other trends may come from an earlier phase of Afro-Asiatic. As I've mentioned before, for instance, a disproportionate number of roots beginning with {d} seem to relate to some sort of destruction, perhaps indicating some sort of long-gone derivation.
* Are there, in these languages, commonly consonants that can appear elsewhere in a prefix, but are dispreferred as the last consonant of a prefix?
* Are there, in these languages, commonly consonants that can appear elsewhere in a suffix, but are dispreferred as the first consonant of a suffix?
* If the answer to either of those last two questions is "yes", are there any commonalities or general tendencies through these languages?
There is such a small set of prefixes that I really don't think it's fair for me to draw conclusions on these questions. And even fewer with multiple consonants in an affix.
* Is there commonly a limit on the length of vowel-clusters?
* If so, what is that limit likely to be?
* Which morphological processes are likely to lead to long vowel-clusters?
For most Semitic languages, the answer is one; vowel clusters are generally not permitted. Hebrew's /ea/ before historical final /X/ and /?\/ (though which has since become phonemic) is an oddity, as most Semitic languages seem to tend toward total assimilation of vowels.

Hebrew and a few other languages may be developing new vowel clusters due to /h/ and /?/ dropping, but I wouldn't call these phonemic yet.

The main process leading to clusters are of course loss of intermediate weak consonants, particularly /h/ and /?/. The Canaanite languages (including Hebrew) show this most prominently, as these consonants were often lost (especially in suffixes), but the resulting vowel clusters quickly monophthongized (compare, for example, Arabic huwa "he" with Hebrew ).

Hebrew /ea ia ua oa/ resulted from a partial lowering of mid-to-high vowels before word-final /X/ and /?\/ (and /h/ in some cases). These then became phonemic as word-final /?\/ and /h/ were lost and /X/ merged with lenited [x] from former /k/ (which did not trigger the lowering).
* What, if any, common types of constraints are likely to be in effect concerning two consonants appearing in a root?
** That is, for example, is it dispreferred to have all three consonants in the root be voiced?
** Or dispreferred for all three to be unvoiced?
** Is it dispreferred to have two of the consonants be the same, either C1C1C2 or C1C2C1 or C2C1C1?
** Is it dispreferred for two of the consonants of the root to be at the same PoA?
** Is it dispreferred for two of the consonants of the root to have the same MoA?
* How about asking the same questions about prefixes instead of roots, or about suffixes instead of roots?
Also an interesting question, but one I can't give you a great deal of information on.

Assimilation in roots, however, is not rare. While the root itself is an abstract concept, the surface realization of conjugated/declined words are subject to the same sorts of allophonic phenomena seen in any other language. Thus, if two root consonants end up in a single cluster, they may well assimilate. If this happens in enough forms of a given root, the assimilation may be generalized to all forms of the root. This is how the actual roots can change slowly over time.

For example, two of the most significant assimilatory phenomena that in many Semitic languages became generalized are voicing and emphasis (ie, pharyngealization). While I can't cite any examples off the top of my head, neighboring consonants (ie, C1 and C2, or C2 and C3) will often acquire the same pharyngealization or voicing state as the other (if possible). However, this type of change to the root itself was by no means universal, even within single languages. The assimilations that occur and then stick seem to be quite sporadic.
* How common are infixes in these languages?
** What are some, if some of them have some?
** Do infixes usually follow the first consonant or precede the last consonant?
** Do infixes usually consist of just one consonant?
Not very common. The only real infix is the -t- of the appropriately-named t-Stem (Hebrew hitpa`el, Arabic 'ifta`ala, etc), which is not universal. This is simply a case of a former prefix undergoing metathesis with the following consonant, either in all cases (as in Arabic) or in certain cases (as in Hebrew, where it only happened before /s š z ṣ/).

The only sort of root-splicing Semitic languages frequently engage in is reduplication. Hebrew, for example, often reduplicates the final two consonants of a root for many different purposes, most commonly diminutives.


If you don't mind my asking, what is your interest in these questions? Most of these just describe phonological trends in Semitic languages, and aren't anything inherent in triconsonantal systems. Or is that just what you were interested in?

I've got to go for now; I'll answer the rest later...

TomHChappell
Avisaru
Avisaru
Posts: 807
Joined: Wed Dec 28, 2005 2:58 pm

Post by TomHChappell »

Mecislau wrote:Sorry for the delay.
No apology necessary! Indeed, let me thank you for your reply.
Mecislau wrote:By "root" I assume you mean "stem", right?
Yes; sorry for the error, thanks for the correction.
Mecislau wrote:(The root is the consonantal base, such as "K-T-B", that is an abstract concept with no surface realization. The stem include vowels that have been filled in between the root consonants, to which affixes are added, such as the Hebrew past tense stem *katav- and future tense stem *-khtov-)
I see.
Mecislau wrote:In that case, no, such vowels are not considered part of the stem.
Thanks. So it's
word = (optional prefix)+stem+(optional suffix)
where
stem = root\template
(should that be root\pattern? or root\transfix?)
, the template being the vowels between the root's first and last consonants and also the reduplication of whichever root consonants get reduplicated (if any).
Mecislau wrote:Hmm. That's an interesting question. I can pretty much safely say that yes, there are definitely some preferences for what consonants can occur where, but I cannot really give you any details. I've never really looked into this that much.
Thanks; that's a solid answer. Do you know of anyone who has looked into it? or anywhere I could look to find whether they looked into it, and what they found if they did?
Mecislau wrote:However, the majority of any such trends are most likely just historical accident. Medial consonantal {w} and {j} are fairly rare, for example, due to the consonants' general weakness and their early merger with the biconsonantal paradigms. Final {h} not resulting from an earlier *w or *j is also very rare if not nonexistent.
Other trends may come from an earlier phase of Afro-Asiatic. As I've mentioned before, for instance, a disproportionate number of roots beginning with {d} seem to relate to some sort of destruction, perhaps indicating some sort of long-gone derivation.
Again, thanks.
Mecislau wrote:There is such a small set of prefixes that I really don't think it's fair for me to draw conclusions on these questions. And even fewer with multiple consonants in an affix.
OK, if I understand (and recall) correctly what you've said both here and before, there are very few suffixes at all, and very few affixes of any type that contain multiple consonants; most affixes are single-consonant prefixes, and there are too few of those to draw any conclusions of the type I was asking about.
But "hit-" is one example of a two-consonant prefix, right?
Mecislau wrote:For most Semitic languages, the answer is one; vowel clusters are generally not permitted. Hebrew's /ea/ before historical final /X/ and /?\/ (though which has since become phonemic) is an oddity, as most Semitic languages seem to tend toward total assimilation of vowels.
Hebrew and a few other languages may be developing new vowel clusters due to /h/ and /?/ dropping, but I wouldn't call these phonemic yet.
The main process leading to clusters are of course loss of intermediate weak consonants, particularly /h/ and /?/. The Canaanite languages (including Hebrew) show this most prominently, as these consonants were often lost (especially in suffixes), but the resulting vowel clusters quickly monophthongized (compare, for example, Arabic huwa "he" with Hebrew ).
Hebrew /ea ia ua oa/ resulted from a partial lowering of mid-to-high vowels before word-final /X/ and /?\/ (and /h/ in some cases). These then became phonemic as word-final /?\/ and /h/ were lost and /X/ merged with lenited [x] from former /k/ (which did not trigger the lowering).
Thanks. That's useful information.
I was asking about "morphological" processes producing vowel-clusters, rather than "phonological" processes (though I suppose any "morpho-phonological processes" that resulted in vowel-clusters would have also been among the things I was asking about).

([EDIT]: Moved, whited-out, and size-zeroed-out some stuff about hypothetical conlangs rather than TRS natlangs. See below. [/EDIT])
Mecislau wrote:
* What, if any, common types of constraints are likely to be in effect concerning two consonants appearing in a root?
Also an interesting question, but one I can't give you a great deal of information on.
Assimilation in roots, however, is not rare. While the root itself is an abstract concept, the surface realization of conjugated/declined words are subject to the same sorts of allophonic phenomena seen in any other language. Thus, if two root consonants end up in a single cluster, they may well assimilate. If this happens in enough forms of a given root, the assimilation may be generalized to all forms of the root. This is how the actual roots can change slowly over time.
For example, two of the most significant assimilatory phenomena that in many Semitic languages became generalized are voicing and emphasis (ie, pharyngealization). While I can't cite any examples off the top of my head, neighboring consonants (ie, C1 and C2, or C2 and C3) will often acquire the same pharyngealization or voicing state as the other (if possible). However, this type of change to the root itself was by no means universal, even within single languages. The assimilations that occur and then stick seem to be quite sporadic.
Thanks.
Mecislau wrote:
* How common are infixes in these languages?
Not very common. The only real infix is the -t- of the appropriately-named t-Stem (Hebrew hitpa`el, Arabic 'ifta`ala, etc), which is not universal. This is simply a case of a former prefix undergoing metathesis with the following consonant, either in all cases (as in Arabic) or in certain cases (as in Hebrew, where it only happened before /s š z ṣ/).
Thanks; I sort of thought that, I guess. I was interested in whether it might be a common or regular thing for the last consonant of a prefix to metathesize with the first consonant of a stem, or for the last consonant of a stem to metathesize with the first consonant of a suffix. Apparently it isn't.
Mecislau wrote:The only sort of root-splicing Semitic languages frequently engage in is reduplication. Hebrew, for example, often reduplicates the final two consonants of a root for many different purposes, most commonly diminutives.
Hmm. I was aware of the gemination of the middle radical for the "intensive" (if that's the right word); I wasn't aware of anything inserting a vowel between two copies of the middle radical, nor of any gemination or reduplication involving the final radical, nor the middle and the final together. Can you exemplify?
(And thanks whether you can or not.)
Mecislau wrote:If you don't mind my asking, what is your interest in these questions? Most of these just describe phonological trends in Semitic languages, and aren't anything inherent in triconsonantal systems. Or is that just what you were interested in?
I was interested not only in what's inherent in TRSes, but also in what can happen in TRSes. Anything that's actually happened in a TRS natlang is relevant for my purpose.
Also, as I explained above, I was interested in "morphological processes" and "morpho-phonological processes", moreso (for that group of questions, at least) than in "purely phonological" processes.
Mecislau wrote:I've got to go for now; I'll answer the rest later...
Thanks, very much, for everything so far!
I'm glad to know you're intending to get around to posting more; I look forward to it.

([EDIT]: Below is some whited-out and size-zeroed-out stuff about hypothetical conlangs rather than TRS natlangs, moved from above. [/EDIT])
[size=0]What I was thinking was something like this:
A template or pattern of vowels (leaving aside for the moment the reduplication of any of the roots in the consonant) could, for some TRS conlang or even some natlang unknown to me, possibly consist of:
* A first vowel V1, and a decision whether it precedes or follows the first radical C1;
* A second vowel V2, and a decision whether it precedes or follows the second radical C2; and,
* A third vowel V3, and a decision whether it precedes or follows the third radical C3.
So if, for example, the root were b-c-d and the vowels of the template were a,e,i, we could still get eight different stems;
* template a-e-i- would create abecid;
* template a-e--i would create abecdi;
* template a--ei- would create abceid;
* template a--e-i would create abcedi;
* template -ae-i- would create baecid;
* template -ae--i would create baecdi;
* template -a-ei- would create baceid;
* template -a-e-i would create bacedi.
Now some of those stems do have leading vowels, some do have trailing vowels, some do have two-vowel internal clusters, and some do have two-consonant internal clusters.

OTOH, a template or pattern of vowels (leaving aside for the moment the reduplication of any of the roots in the consonant) could, for some TRS conlang or even some natlang unknown to me, possibly consist of up to six vowels-or-absences-of-vowels:
* A vowel U1, to precede the first radical C1 (or a decision that there be no such vowel);
* A vowel V1, to follow the first radical C1 (or a decision that there be no such vowel);
* A vowel U2, to precede the second radical C2 (or a decision that there be no such vowel);
* A vowel V2, to follow the second radical C2 (or a decision that there be no such vowel);
* A vowel U3, to precede the third radical C3 (or a decision that there be no such vowel);
* A vowel V3, to follow the third radical C3 (or a decision that there be no such vowel).
If the language has three vowels /a,i,u/,
there are (3+1)^6 = 4,096 possible templates; but of course they're not really all distinct since, for instance, it'd be impossible to distinguish
/abacada/ where V1 and V2 are /a/ but U2 and U3 are absent, from /abacada/ where V1 and V2 are absent but U2 and U3 are /a/.
Also I think such a language (if it existed) probably wouldn't allow three-consonant clusters in the stem, so you couldn't have templates with all four of V1, U2, V2, and U3 be "absent". In fact it may not allow two-consonant clusters; or it may not allow any three consecutive vowels of the template to be absent (if that makes any sense?)

If a stem could start with a vowel and a prefix could end with a vowel, you might get a vowel-cluster; if a stem could end with a vowel and a suffix could begin with a vowel, you might get a vowel-cluster; if a compound word had a vowel-final stem followed immediately by a vowel-inital stem, you might get a vowel-cluster.

Those were the kind of "morphological processes leading to vowel-clusters" I was asking about. You've said they're quite unlikely in Semitic languages, because stems are almost never vowel-intial or vowel-final, and stems usually have no internal vowel clusters; in particular templates or patterns never (? or almost never?) put vowel-clusters into a triconsonantal stem.

But, if I remember a PM from you and understood it correctly, some have suggested that biconsonantal roots can for some purposes conveniently be considerd triliteral roots whose middle letter is a vowel instead of a consonant; if that's so a regular template applied to such a root could result in a three-vowel internal cluster in the stem, couldn't it?

However in Semitic languages vowel-clusters almost never arise through regular morphology (if I'm understanding you); there are certain phonological processes that can result in vowel-clusters, but they're almost purely phonological, having little or no "morphological character".[/size]

User avatar
Wiseblood
Sanci
Sanci
Posts: 22
Joined: Thu Jul 19, 2007 8:46 pm
Location: Canada

Post by Wiseblood »

Hi all. Sorry Tom, I got your PM some time ago, but Real Life TM interfered with me coming here sooner. There is so much written now that I'll probably need to re-read everything a few times to fully form some useful answers.

In any case, I figure I'll just go through and touch on some points that strike me in the meantime.

Sugar/diabetes - the root S-K-R describes intoxication/alcoholism in Arabic - سكرانُ /sakrānu/ "alcoholic", سكِر /sakira/ "he got wasted", etc. Although, now that I look at it, I'm pretty sure that the Akkadian word is /šakrānu/, so the roots aren't cognates. ;-(

The root S-F-Ḥ is still productive in Arabic, although what the semantic connexion to the housefold/family is is completely beyond me. Off the top of my head are the verb سفح /safaḥa/ "he shed blood", and the "noun of profession" سفّاح /saffāḥun/ which means something like "slasher" or "butcher" (in the psycho-killer sense). While I'm pretty sure the word مسفحة /misfaḥatun/ doesn't exist in Arabic, if it did it would mean something like "blood-letting instrument", or perhaps more loosely, "fluid letting device". Actually, if we look at "fluid letting device" with junior-high eyes (pointed straight at the gutter), perhaps the connexion to family (or at least reproduction) becomes more obvious. ;-)
TomHChappell wrote:How do you form the intensive of a tetra-consonantal root?
Hmmm, usually you can't. However, Arabic does have special derivative forms for quadra-consonantal roots that can alter the valiency of the verb, and these differ somewhat from those used for tri-consontal verbs. I should also note that quadra-consontals that are really doubled bi-consonants like زلزل /zalzala/ "shake" and قهقه /qahqaha/ "laugh boisterously" are considered frequentatives/intensives by their nature.
TomHChappell wrote:If there are pre-affricated or post-affricated or pre-aspirated or post-aspirated or pre-nasalized or post-nasalized consonant phonemes in an Afro-Asiatic Tri-Consonantal-Root-System language, how do speakers and hearers distinguish the "complexconsonant phonemes" from a consonant-cluster?
PAA is usually reconstructed as having at least a half-dozen affricates along with a similar number of labialised stops. PS itself probably had /ts/,/dz/,/ṭṣ/,/tɬ/ and /ṭɬ/; these became /s/,/z/,/ṣ/,/š/ and /ḍ/ in Arabic (altough the latter two were still laterals in the middle ages) and /s/,/z/,/ts/,/s/ and /ts/ in MH, respectively (there's some good evidence that many of these were still affricates in biblical times though). Many modern Arabic dialects have /č/ and/or /ts/ as well. Ambiguity seems to be avoided by negating sequences that resemble the affricates - ie the sequences /t+s/ and /t+š/ simply don't occur in Arabic.

There's actually a fair bit of research out there on phonotactic restrictions in PAA and Semitic roots. PAA and PS definitely had restrictions on the kinds/sequences of phonemes that were allowed in tri-consonantals.
TomHChappell wrote:What about compounds, e.g. compound nouns?
Mecislau (as always) covered this pretty well. I can confirm that our strategies in Arabic are pretty similar to those in Hebrew:

ابنُ آدمَ /ibn.u-ʔādam.a/ "human being" (lit. "son of Adam"): son.CONST.NOM.SG-Adam.ABSOL.GEN.SG
بنو آدمَ /ban.ū-ʔādam.a/ "human beings" (lit. "sons of Adam"): son.CONST.NOM.PL-Adam.ABSOL.GEN.SG

محطّة قطارٍ /maḥaṭṭ.at.u-qitār.in/ "train station" station.FEM.CONST.NOM.SG-train.ABSOL.GEN.SG
محطّة القطار /maḥaṭṭ.at.u-l-qitār.i/ "the train station" station.FEM.CONST.NOM.SG-the-train.DEF.GEN.SG

There are a *really* small number of IE-style compounds in Arabic, and the only one I can think of now is رأسماليّة /raʔsamāliyyatun/ "capitalism" from رأس /raʔsun/ "head" + مال /mālun/ "wealth". There is also something called an "absolute compound" that consists of two nouns in the definite accusative placed next to each other. This is mostly restricted to place names بيتَ لحمَ /bayt.a lam.a/ "Bethlehem": house.DEF.ACC.SG meat.DEF.ACC.SG and numbers in the 'teens ثلاثةَ عشرَ /θalāθ.at.a ʕašar.a/ "thirteen": three.FEM.DEF.ACC.SG ten.DEF.ACC.SG.
Mecislau wrote:Well, first of all, the triconsonantal system is only found in Semitic, not all of Afro-Asiatic, and Semitic has much less of that.
Okay, I really have to ask where you're getting your information here (see above about affricates). While it's true that Semitic (and perhaps Arabic in particular) take the whole "templative" morphology thing to an extreme, and that vowels seem pretty stable in the "southern" branches of AA (Chadic and Cushitic), I think it is pretty uncontroversial that at least Berber and Ancient Egyptian both have/had similar derivational systems based on tri-consonantal roots. The system isn't quite as productive as it once was in Berber due to massive vowel reductions that occured at some point, but it is most certainly still there. In Egyptian's case, I know for a fact that not only did it have templative morphology, but many of the forms correspond to those in Semitic. I can give examples, if you'd like. Also, there is some research to indicate that the largly bi-consonantal nature of Cushitic and Chadic are due to consonant loss and that many of their roots were tri-consonantal originally (altough I'm pretty sure the stability of vowels is still in contention).
Mecislau wrote:For example, the root *Q-R-' "read, call, invite" has two passive participles in pa`al: the expected קרוא qaru' "invited", and the unexpected קרוי qaruy "named" that would be expected from a hypothetical root *Q-R-H. On the other hand, the real root *Q-R-H "happen, occur" has two different forms of the feminine singular of its active participle existing in free variation, the expected קורה qorah and the unexpected קורית qoret which was acquired from the final *' paradigm.
Interesting, the passive participles for these verbs in Arabic are مقروء /maqrū'un/ and مقريّ /maqriyyun/ respectively, and the active participles are قارئ /qāriʔun/ and قارٍ /qārin/ respectively, so I guess there was less chance for ambiguity to happen there. One thing though, are you sure the root would be Q-R-H and not Q-R-Y? I ask only because the verb قرى /qarā/ means "invite" or "show hospitality" in Arabic, and that has a pretty clear connexion to words like قرية /qaryatun/ "village". Actually, now that I think about it, I'm pretty certain that /qaryat/ was the Phoenicion word for "city". What's the Hebrew word?
TomHChappell wrote:Unless all those guys I just quoted are wrong? I have no way of telling how reliable they are.
Again, my experience is certainly contrary to Mecislau's here.
TomHChappell wrote:MASSIVE AMOUNTS OF QUESTIONS
Good God! I think I'll need some time to go through these. O_o
Mecislau wrote:That is, Semitic has *K-T-B, while Berber (for example) has ktVb - three consonants, but not all unlinked from one another.
Did you check the link he provided? The data is pretty sparse, but I spotted at least once example of a stem vowel changing: ad elmedæɣ Tæmasæq "I will learn Tamasheq" vs. lammædæɣ Tæmasæq "I am learning Tamasheq". I know I've seen data from Berber dialects where each of five different tenses will have different vowels for the same verb. This isn't even getting into the nominal morphology - many Berber words have "broken plurals" similar to those in Arabic.

Even branches like Beja with so-called "stable" vowels show vowel alternations/resyllabification. Although, whether or not that might better be described as "ablaut" or somesuch is debatable. Again, I'll fish out some data, if anyone wants some.
Mecislau wrote:That's... not entirely accurate. Sure, Arabic verbs may be disyllabic in MSA, but that's just because MSA always has non-zero verbal suffixes (IIRC). Modern Arabic dialects, though, frequently are only a single syllable. Glides are also acceptable parts of roots, but due to historical sound changes they tend to be very "weak" and prone to dropping, unless they were geminated. Most of the "irregular" paradigms in modern Semitic languages were at one time regular.
Is there a usage of disyllabic I don't know about? I'm seriously asking this because I thought it just meant that a word had to have two syllables (I *really* don't mean this in a sarcastic way - I know I've been coming off like a jerk, but I'm really curious about all of this stuff). Or are you reading that as the *stem* has to be disyllabic? For the record, the jussive actually has a zero ending in Classical Arabic. I really don't know what you mean by "single syllable" here. However, I absolutely agree that describing verbs with "weak" radicals as irregular is probably not a good idea.

Okay, that's all I have time for now, I'll try to get back to this sooner.

User avatar
Mecislau
Avisaru
Avisaru
Posts: 491
Joined: Thu Jul 24, 2003 2:40 pm
Location: Maryland
Contact:

Post by Mecislau »

My apologies for any inaccuracies or mistakes I've made in the descriptions of languages I'm less familiar with. My familiarity with Semitic and Afro-Asiatic outside of Hebrew is relatively limited, and I hope I was clear in saying so. What I was talking about with regards to Berber is drawing on memory of some Tamazight resources that I read a long time ago, so if my memory has failed me, I'm sorry. I just seem to remember that the verbal system seemed to significantly favor simple ablaut involving a single vowel rather than the complex sort of Semitic languages. Of course, that's just a single Berber language I've read about, and since it's true that there has been a great deal of vowel reduction, I have no idea what has been lost.

That said, it is good to see you again, Wiseblood! You're certainly a lot more knowledgeable about Semitic and Afro-Asiatic languages than I am, and I always enjoy reading whatever information you provide.
Wiseblood wrote:Sugar/diabetes - the root S-K-R describes intoxication/alcoholism in Arabic - سكرانُ /sakrānu/ "alcoholic", سكِر /sakira/ "he got wasted", etc. Although, now that I look at it, I'm pretty sure that the Akkadian word is /šakrānu/, so the roots aren't cognates. ;-(
Not likely. Hebrew has שיכור šikur for "drunk", while the root for "sugar" (and "diabetes") has /s/, not /š/.
Wiseblood wrote:Interesting, the passive participles for these verbs in Arabic are مقروء /maqrū'un/ and مقريّ /maqriyyun/ respectively, and the active participles are قارئ /qāriʔun/ and قارٍ /qārin/ respectively, so I guess there was less chance for ambiguity to happen there. One thing though, are you sure the root would be Q-R-H and not Q-R-Y? I ask only because the verb قرى /qarā/ means "invite" or "show hospitality" in Arabic, and that has a pretty clear connexion to words like قرية /qaryatun/ "village". Actually, now that I think about it, I'm pretty certain that /qaryat/ was the Phoenicion word for "city". What's the Hebrew word?
Well, you're also correct. This is a slightly messy case where "traditional" grammatical analysis and "modern" analysis use different schemes to represent the same thing. The root *Q-R-H is the traditional representation of what modern linguists would call *Q-R-Y. I'm guessing this is a result of the citation form (the usual 3SgMasc perfective) ending in orthographic H in Hebrew (קרה qarah /kara/ "it happened").

As far as I'm aware Hebrew lacks a cognate to the Phoenician "city"/Arabic "village".
Wiseblood wrote:Is there a usage of disyllabic I don't know about? I'm seriously asking this because I thought it just meant that a word had to have two syllables (I *really* don't mean this in a sarcastic way - I know I've been coming off like a jerk, but I'm really curious about all of this stuff). Or are you reading that as the *stem* has to be disyllabic? For the record, the jussive actually has a zero ending in Classical Arabic. I really don't know what you mean by "single syllable" here. However, I absolutely agree that describing verbs with "weak" radicals as irregular is probably not a good idea.
... you know, I have no idea what I was trying to say there :? That's rather embarrassing.

But I'm not sure how relevant it would've been in the first place, since I clearly misunderstood the intent of the quoted text anyways.

hwhatting
Smeric
Smeric
Posts: 2315
Joined: Fri Sep 13, 2002 2:49 am
Location: Bonn, Germany

Post by hwhatting »

Mecislau wrote:
Wiseblood wrote: I ask only because the verb قرى /qarā/ means "invite" or "show hospitality" in Arabic, and that has a pretty clear connexion to words like قرية /qaryatun/ "village". Actually, now that I think about it, I'm pretty certain that /qaryat/ was the Phoenicion word for "city". What's the Hebrew word?
As far as I'm aware Hebrew lacks a cognate to the Phoenician "city"/Arabic "village".
Sorry if this is way off, but what about qiryat as in Qiryat Shmona etc.? Is that an Aramaic loan or does it belong to a different root?

Post Reply