The short answer is the accent. A major weakness of the earlier version
of my model is that I wanted to derive accent secondarily. But this
caused everything to be too complex and I could not get the rules to
I am now looking more into how PU corresponds with PIE. And it is like
exploring a pristine land. I just look at the data, let my mind work,
take notes and be totally amazed at what I find.
Here is an overview of the current model.
And also something I wrote this week in my notes about the PIU numerals.
Disclaimer: I am not a professional linguist. What I write reflects ideas
of mine. These ideas are very speculative.
My Vowel Harmony to Ablaut Model
1. Pre-PIE0 vowel system
PIU **ä merges with Pre-PIE0 **e and sometimes even **i
PIU **ë shifts from unrounded back to rounded front **ö
Pre-PIE0 ï (-> PIE eu) develops secondarily through vowel harmony
Close vowels: i~ü (ï)-u
Open vowels: e~ö~a~o
The vowel harmony becomes a system that looks a lot like the Turkic vowel cube.
Close vowels have rounding harmony.
Open vowels have both front/back and rounding harmony.
2. Verbal prefixes
PIE had the following ways to prefix a verb:
1. Reduplication: DvCvC
Remnants of this have been preserved in many branches
2. Single vowel prefix: vCvC
Remnants of this have been preserved in the Greek/Indo-Iranian augment
The Pre-PIE1 verb always had a prefix. These prefixes + vowel harmony cause
verbs to be in either o-grade or e-grade depending on conjugation.
3. Vowel collapse
For verbs, vowel harmony caused the following vowel system in the root
Close vowels: i ï
Open vowels: e (öh1, ah2, oh3)
Close vowels: ü u
Open vowels: o (oH)
For nouns, ö and a were also removed, but could be kept with a laryngeal.
The old root/initial syllable accent shifted mostly to the modern PIE accent
5. E/O shift
O-grade words get reworked.
All o-grade syllables but one are converted to e-grade.
When stress is on the initial syllable (acrostatic nouns) the remaining o-grade syllable
will go to the end of the word.
When stress is on any other syllable (root/ending in stative verbs) the remaining o-grade
syllable will be in the root.
Accented close vowels split
'i -> éi
'ü -> ói
'ï -> éu
'u -> óu
Unaccented close vowels merge
i/ü -> i
ï/u -> u
Open vowels get reduced to zero grade in unaccented positions except for:
- O-grade words with initial or root stress.
- The vowel in a reduplication prefix
O-Acrostatic *genh1os - tribe
nom/acc sg. 'gonhos -> genhos
gen sg. 'gonhosos -> genhesos
O-Amphikinetic *leymon - lake
nom sg. 'lümons -> 'limons
gen sg. lümon'os -> lümen'es -> limn'es
Stative *memone - to think, to be mindful to remember
3rd person sg. (m)o'monü -> (m)e'moni (later me'mone)
3rd person pl. (m)omon'or -> (m)emon'er -> memn'er
It has often been speculated that the IE words for 4 (*kʷetwores) and 8 (*Hoḱtṓw)
are related. If that is true, they should come from an earlier form like **kʷoktw.
In the word for 4 the 't' was elided, and a suffix was added.
In the word for 8 the inital kʷ was lenited to a laryngeal.
Several etymologies have been proposed for this. The usual etymology for 8 relates
it to an old Avestan word ašti- 'breadth of four fingers'. *Hoḱtṓw would then be
a dual of that. And the word for 4 would have been the original singular.
But when we then look at the Uralic side, another cognate emerges.
Their word for the number 2 is reconstructed as käktä/kakta. This looks a lot
like our **kʷoktw. But now the meaning of '4 fingers' does not hold up.
And we get down to an original meaning of 'pairing', which has also
But even that meaning may be too much. For if we look at the reconstructed
Uralic numbers, we find more forms that look eerily alike:
PU: *(w)ük(t)i one (w reflected in Mordvinic)
PU: *käktä two
PU: *witte five
PU: *kutte six
Now, the Uralic word *ük(t)i can be tied to the IE word for one. The Indo-Iranian
languages have a form 'eka' that derives from *Hoykos. The rest of the IE languages
use the form *HoyH-nos
And on the IE side, I should also add the word (s)wéḱs 'six' to this paradigm.
This can be derived from earlier **kukte via ukte -> ueks.
Alternatively, it can also be derived from PU *witte, earlier **wikte -> weks
with a change in meaning.
The 's' in the word (s)weks was later added by analogy with the Semitic word for six.
This gives me the following numbers on the IE side:
PIE: *Hóynos / *HóyHnos
(And it is even possible that the IE word for 5 *pénkʷe fits into this paradigm.
Italo-Celtic had an alternative form *kʷenkʷe/*kʷinkʷe. But there are problems
with this. Most of IE reflects forms with a p-, not kʷ-. And the 'n' in the middle
would need explanation.
Also, the IE word for 2 *dwóh1 could be derived with a sequence
like kaktwo -> agdwo -> dwo. But the complete absence of any sign of the
original first syllable makes me very reluctant here.)
Could these all numbers really come from one single root? It seems unlikely.
But what if?
1. What would such a root mean?
Well, at least the meaning of 'pairing' will not hold up with odd numbers like 1 and 5.
My first thought would be something like 'number' or 'count'.
2. Is it possible to find a possibly cognate word with such a meaning (number/count)
in either IE or Uralic?
First the IE side: In English we have the word 'quota'. This is a borrowing
from Latin and has meanings such as 'maximum number' in English. The Latin word
ultimately derives from a Latin interrogative 'quot', which means 'how much/
what number'. The reconstructed PIE form for this word is:
PIE *kʷoti how much?
Then the Uralic side. There we have this root:
PU *kokɜ - size
So it is plausible that these words trace back to an earlier form like **kʷokti.
If this is all correct, then a very primitive paradigm emerges. In the earliest
stage, one word was used to indicate a number. That word meant as much as 'this much'.
One can imagine that a speaker would hold up his/her hands with the appropriate
number of fingers when using this word. At a later stage, they started to dissimilate
this word. For every number another variant was used. Also, they replaced some
numbers with alternative words outside of this paradigm.
But I'm not done, yet. That is because there is a rough sound law hidden in this data.
One thing that I note is that the only thing that consistently changes between the
number words in PU is the vowel. And the k/w alternation at the start of the word
is an almost regular sound change.
Let me summarize this:
Form PIU form PU form PU reflex of **kʷ- PU vowel
Interrogative: **kʷokt- *koke k o
Number 1: **kʷykt- *(w)ük(t)e w->none ü
Number 2: **kʷakt- *kakta k ä/a
Number 5: **kʷikt- *witte w i
Number 6: **kʷukt- *kutte k u
But before I phrase the sound change, I must look at an alternative reflex of PIU **kʷ
that is found in the interrogative pronouns:
PU *ke /*ku- ~ *ko- PIE *kʷo- who
PU *mi PIE *kʷi- what
Here we see 'm-' as a reflex of PIU **kʷ before i.
So I can now formulate a rough sound law.
- PIU **kʷ regularly changes to PU *w or *m before i/ü,
(and this also happens incidentally before e/ä)
otherwise it changes to PU k
And this is also valid for other PIE labiovelars than *kʷ
PU: *mińä daughter-in-law, young woman
PIE: *gʷen- woman
PU: *weδɜ- to kill
PIE: *gʷʰen- to strike, slay, kill
PU: *wäke- 'come; go, run'
PIE: *gʷeh2- (perfective) to go
PU: *lewe- 'throw, shoot (tr)'
PIE: *gʷelH- 'to throw, reach, pierce; to hit by throwing'