Who’s who in the Zagi family?

Using only the internal structure of the Zagi tablets, without guessing sounds or meanings, I argue that the Kristiansen glyphs record a fully consistent family tree for the “clan of Zagi,” complete with parents, children, grandparents, and even a couple of probable scribal slips.

In my previous article I argued for adding the following to the Klema dictionary:

• A02-L01 (previously “AND_PLUS”): AND
• C03-M03-T02 (Üksküla’s “def_pivot”): HAVE, MAKE, BEGET
• C03-M03-L01 (Üksküla’s “eq_pivot”): BE, =
• C01-M03-S01-C02: CHILD

That gives us, from what Üksküla previously called “X DEF_PIVOT N CHILD” lines:

• F3_44: C02-S01-M03-S01 HAVE TWO CHILD
• F3_45: C05-C02-T02-M03-T03 HAVE TWO CHILD
• F3_46: C05-C02-T02-M03-T03 HAVE TWO CHILD
• F3_51: B03-T02-B05-L01 HAVE TWO CHILD
• F3_52: A01-L01-B03-T02 HAVE TWO CHILD
• F3_53: A01-L01-B03-T02 HAVE TWO CHILD
• F3_58: C02-S01-M03-S01 HAVE TWO CHILD
• F3_59: C05-C02-T02-M03-T03 HAVE TWO CHILD
• F3_60: C05-C02-T02-M03-T03 HAVE TWO CHILD
• F3_70: C02-S01-M03-S01 HAVE THREE CHILD
• F3_71: C06-S01-C02-S01-C01-T02 HAVE THREE CHILD
• F3_72: C06-S01-C02-S01-C01-T02 HAVE THREE CHILD
• F4_83: C07-S02-C06 HAVE FOUR CHILD

…and from “CHILD N EQ_PIVOT Y” lines:

• F3_47: CHILD ONE = B03-T02-B05-L01
• F3_48: CHILD TWO = C03-T03-M01-T02
• F3_54: CHILD ONE = C07-S02-C06
• F3_55: CHILD TWO = B01-L01-B02-S01
• F3_60: CHILD ONE = B03-T02-B05-L01
• F3_61: CHILD TWO = C03-T03-M01-T02
• F3_72: CHILD ONE = A01-L01-B03-T02
• F3_73: CHILD TWO = T03-M01-T02
• F3_74: CHILD THREE = A03-L01-A04-T03
• F4_84: CHILD ONE = C07-S01-A05-S01
• F4_85: CHILD TWO = C05-C01-T02-C06-T02
• F4_86: CHILD THREE = A04-T02-A01-T02
• F4_87: CHILD FOUR = M01-T02-B02-T03

…and lastly parental couples (X AND_PLUS Y) in def_pivot CHILD lines:

• F3_46: C02-S01-M03-S01 AND C05-C02-T02-M03-T03 HAVE TWO CHILD
• F3_53: B03-T02-B05-L01-S01 AND A01-L01-B03-T02 HAVE TWO CHILD
• F3_60: C02-S01-M03-S01 AND C05-C02-T02-M03-T03 HAVE TWO CHILD
• F3_72: C02-S01-M03-S01 AND C06-S01-C02-S01-C01-T02 HAVE THREE CHILD

We have no idea how the codes translate to sounds. We do now what codes are very probably vowels or consonants. I didn’t feel like continuing to use codes, so I wrote a short script to help me create a somewhat plausible transliteration.

I started by adding vowels in rough order of frequency (there are eleven vowels, I didn’t want too many accented letters in there):

I then had the script distribute random consonants in rough order of frequency and shuffled them around until I got a text where most words are at least pronounceable:

I attach no real meaning or sounds to the letters, only keeping to the convention that one Kristiansen glyph = 1 sound. My instincitive reasoning behind this is that the average word length is too short to warrant a lot of either vowel or consonant digraphs.

While there’s not a chance that my transliteration even gets anywhere close to the ground truth, this does give my pattern-recognising eye something easier than a list of either too-similar Kristiansen codes or unreadable Kristiansel glyphs.

That said, I get the following for the results above (with duplicate lines removed)

…which sounds reasonable. Document 4 looked like a nice big family; transliterating the entire document gives me this, which already sounds like something plausible to my ears:

séš pnê knéšê tnul . tnul tùpa pni subu . tnul lzâre pni ktêšê . tnul fzôžo pni yêwê . tnul knéšê pni rêxa . tnupém šu séš pni subu . tnupôv šu séš pni ktêšê . tnupôv šu séš pni yêwê . tnupém šu séš pni rêxa . mašôv šu subu pni séš . mašôv šu ktêšê pni séš . mašôv šu yêwê pni séš . mašôv šu rêxa pni séš . subu pni tnupém šu séš . subu pni tnupém šu tnupôv šu mêgi . subu pni tnupém šu tnupôv šu tnupôv šu lunu . subu pni tnupém šu tnupôv šu tnupôv šu klêna . subu pni tnupém šu tnupôv šu wimê . subu pni tnupém šu tnupôv šu tnupém šu lunu . subu pni tnupém šu tnupôv šu tnupém šu šulutê . ka pni séš . tnav pni ma šu séš . .

Now splitting this up in a line per sentence and replacing the words we most probably know with English translations, we get:

séš HAVE FOUR CHILD.
CHILD ONE BE subu.
CHILD TWO BE ktêšê.
CHILD THREE BE yêwê.
CHILD FOUR BE rêxa.
tnupém šu séš BE subu.
tnupôv šu séš BE ktêšê.
tnupôv šu séš BE yêwê.
tnupém šu séš BE rêxa.
mašôv šu subu BE séš.
mašôv šu ktêšê BE séš.
mašôv šu yêwê BE séš.
mašôv šu rêxa BE séš.
subu BE tnupém šu séš.
subu BE tnupém šu tnupôv šu mêgi.
subu BE tnupém šu tnupôv šu tnupôv šu lunu.
subu BE tnupém šu tnupôv šu tnupôv šu klêna.
subu BE tnupém šu tnupôv šu wimê.
subu BE tnupém šu tnupôv šu tnupém šu lunu.
subu BE tnupém šu tnupôv šu tnupém šu šulutê.
ka BE séš.
tnav BE ma šu séš.

The first five lines are transparent:

I’ll return to the rest later. Document 1 and 2 don’t have any child or parent information at first sight. Document 3 has this:

lunu HAVE TWO CHILD.
klêna HAVE TWO CHILD.
lunu AND klêna HAVE TWO CHILD.
CHILD ONE BE mêgi.
CHILD TWO BE parê.
tnupôv šu lunu AND klêna BE mêgi.
tnupôv šu lunu AND klêna BE parê.
mêgi HAVE TWO CHILD.
wimê HAVE TWO CHILD.
mêgiu AND wimê HAVE TWO CHILD.
CHILD ONE BE séš.
CHILD TWO BE fixu.
tnupôv šu mêgi AND klêna BE séš.
tnupôv šu mêgi AND klêna BE fixu.
lunu HAVE TWO CHILD.
klêna HAVE TWO CHILD.
lunu AND klêna HAVE TWO CHILD CHILD ONE BE mêgi.
CHILD TWO BE parê.
tnupôv šu lunu AND klêna BE mêgi.
tnupôv šu lunu AND klêna BE parê.
mêgi BE tnupôv šu tnupôv šu lunu AND klêna.
tnupôv šu mêgi BE séš.
tnupôv šu mêgi BE fixu.
parê BE tnupôv šu tnupôv šu lunu AND klêna.
tnupôv šu parê BE zêt.
tnupôv šu parê BE néz.
lunu HAVE THREE CHILD.
šulutê HAVE THREE CHILD.
lunu AND šulutê HAVE THREE CHILD CHILD ONE BE wimê.
CHILD TWO BE arê.
CHILD THREE BE žiya.
tnupém šu lunu AND šulutê BE wimê.
tnupôv šu wimê BE séš.
tnupôv šu wimê BE fixu.
tnupém šu lunu AND šulutê BE arê.
tnupôv šu lunu AND šulutê BE žiya.
tnupôv šu žiya BE klêv.
tnupém šu žiya BE rêxa.
tnupôv šu žiya BE fépéi.

There’s some complete duplication in there, but these are the families we get:

Disregarding duplicates, repetition and lines that do not contain CHILD I read this, in order:

lunu AND klêna HAVE TWO CHILD.
CHILD ONE BE mêgi.
CHILD TWO BE parê.
mêgiu AND wimê HAVE TWO CHILD.
CHILD ONE BE séš.
CHILD TWO BE fixu.

lunu AND šulutê HAVE THREE CHILD
CHILD ONE BE wimê.
CHILD TWO BE arê.
CHILD THREE BE žiya.

I see this as two sets fo grandparents, and I don’t think it’s a huge leap to tentatively map it out like this (the only assumption being that the two “lunu” are actually two different people, lunu₁ and lunu₂):

Turning to document 1, I get this transcription:

séš BE xzor.
séš BE xzorôv.
séš BE mašôv.
séš BE tnupôv.
mêgi BE xzor.
mêgi BE xzorôv.
mêgi BE mašôv.
mêgi BE tnupôv.
mêgi BE mašôv šu séš.
mašôv šu séš BE mêgi.
tnupôv šu mêgi BE séš.
séš BE tnupôv šu mêgi.
lunu BE mašôv šu mêgi.
mašôv šu mêgi BE lunu.
tnupôv šu lunu BE mêgi.
mêgi BE tnupôv šu lunu.
klêna BE mašém šu mêgi.
mašém šu mêgi BE klêna.
tnupôv šu klêna BE mêgi.
mêgi BE tnupôv šu klêna.
lunu AND klêna BE mah šu mêgi.
mah šu mêgi BE lunu AND klêna.
séš BE tnupôv šu tnupôv šu lunu.
séš BE tnupôv šu tnupôv šu klêna.

Any person in a family tree is mlost probably either a man or a woman (I know, I know). That also means that they are a child (son or daughter), a parent (father or mother), a grandchild (grandson or granddaughter), a sibling (brother or sister), a nephew or niece, etc.

If these documents are an actual primer like Rubergskier’s “Dozenal primer” we could assume these types of relationships would be clarified — and I think this is what the start of Document 1 does. We’re told who/what Séš is:

• xzor, but also xzorôv
• mašôv
• tnupôv

Mêgi, who we can assume is a parent of Séš, is also “xzor” and “xzorôv”, “mašôv” and “tnupôv”. We explitly get told both

• mêgi BE mašôv šu séš / mašôv šu séš BE mêgi
• séš BE tnupôv šu mêgi / tnupôv šu mêgi BE séš

I see no other way to reasonably decode this than

• mêgi BE PARENT šu séš / PARENT šu séš BE mêgi
• séš BE CHILD šu mêgi / CHILD šu mêgi BE séš

This neatly shows the commuitative property of “BE”: A is B and B is A. It’s not a huge leap to read the “šu” in phrases like “A is parent šu B” and “B is child šu A” as simply “of”. Taking mašôv=PARENT, tnupôv=CHILD and šu=OF, Document 1 gives us the following information:

The transliteration makes it clear that even in this short document we get a lot of almost-duplicate words:

• xzor / xzor-ôv
• mah / maš-ôv / maš-ém

Looking at the shorter Document2 we get less redundancy, but the same patterns emerge:

This gets us:

• xzor / xzor-ôv
• mah / maš-ôv / maš-ém
• tnul / tnup-ôv / tnup-ém

“Lunu and šulutê are mah of wimê” tells me “mah” is gender neutral PARENT, just like “tnul” is gender neutral CHILD. “Lunu is mašôv of wimê” together with “šulutê is mašém of wimê” indicates that “-ôv” and “-ém” indicate a gender (there is no way to find out which is which).

I translate “tnupôv” as CHILD_A and “tnupém” as CHILD_B, which allows me to now fully translate Document 3:

lunu HAVE TWO CHILD.
klêna HAVE TWO CHILD.
lunu AND klêna HAVE TWO CHILD.
CHILD ONE BE mêgi.
CHILD TWO BE parê.
CHILDA OF lunu AND klêna BE mêgi.
CHILDA OF lunu AND klêna BE parê.
mêgi HAVE TWO CHILD.
wimê HAVE TWO CHILD.
mêgiu AND wimê HAVE TWO CHILD.
CHILD ONE BE séš.
CHILD TWO BE fixu.
CHILD_A OF mêgi AND klêna BE séš.
CHILD_A OF mêgi AND klêna BE fixu.
lunu HAVE TWO CHILD.
klêna HAVE TWO CHILD.
lunu AND klêna HAVE TWO CHILD
CHILD ONE BE mêgi.
CHILD TWO BE parê.
CHILD_A OF lunu AND klêna BE mêgi.
CHILD_A OF lunu AND klêna BE parê.
mêgi BE CHILD_A OF CHILD_A OF lunu AND klêna. ???
CHILD_A OF mêgi BE séš.
CHILD_A OF mêgi BE fixu.
parê BE CHILD_A OF CHILD_A OF lunu AND klêna. ???
CHILD_A OF parê BE zêt.
CHILD_A OF parê BE néz.
lunu HAVE THREE CHILD.
šulutê HAVE THREE CHILD.
lunu AND šulutê HAVE THREE CHILD
CHILD ONE BE wimê.
CHILD TWO BE arê.
CHILD THREE BE žiya.
CHILD_B OF lunu AND šulutê BE wimê.
CHILD_A OF wimê BE séš.
CHILD_A OF wimê BE fixu.
CHILD_B OF lunu AND šulutê BE arê.
CHILD_A OF lunu AND šulutê BE žiya.
CHILD_A OF žiya BE klêv.
CHILD_B OF žiya BE rêxa.
CHILD_A OF žiya BE fépéi.

This is reinforced in Document 4, which can now also be almost fully translated:

séš HAVE FOUR CHILD.
CHILD ONE BE subu.
CHILD TWO BE ktêšê.
CHILD THREE BE yêwê.
CHILD FOUR BE rêxa.
CHILD_A OF séš BE subu.
CHILD_B OF séš BE ktêšê.
CHILD_B OF séš BE yêwê.
CHILD_A OF séš BE rêxa.
mašôv OF subu BE séš.
mašôv OF ktêšê BE séš.
mašôv OF yêwê BE séš.
mašôv OF rêxa BE séš.
subu BE CHILD_A OF séš.
subu BE CHILD_A OF CHILD_B OF mêgi.
subu BE CHILD_A OF CHILD_B OF CHILD_B OF lunu.
subu BE CHILD_A OF CHILD_B OF CHILD_B OF klêna.
subu BE CHILD_A OF CHILD_B OF wimê.
subu BE CHILD_A OF CHILD_B OF CHILD_A OF lunu.
subu BE CHILD_A OF CHILD_B OF CHILD_A OF šulutê.
ka BE séš.
tnav BE ma OF séš.

Two lines in Document 3 seem inconsistent:

• mêgi BE CHILD_A OF CHILD_A OF lunu AND klêna.
• parê BE CHILD_A OF CHILD_A OF lunu AND klêna.

Read literally, these lines make Mêgi and Parê grandchildren of Lunu and Klêna. I consider this very likely a scribal slip, because everywhere else in all four documents CHILD / CHILD_A / CHILD_B behave as straightforward “child”, never as “descendant” or any more abstract relation.

In particular, we have these statements elsewhere:

• Document 3:
  • lunu AND klêna HAVE TWO CHILD.
  • CHILD ONE BE mêgi. CHILD TWO BE parê.
• Document 4:
  • subu BE CHILD_A OF CHILD_B OF mêgi (i.e. subu, child of séš, is the grandchild of mêgi)
  • subu BE CHILD_A OF CHILD_B OF CHILD_B OF lunu.
    subu BE CHILD_A OF CHILD_B OF CHILD_B OF klêna.

Taken together, these lines consistently place Mêgi and Parê as children (not grandchildren) of Lunu and Klêna, and Subu as their grandchild. To rescue the “CHILD_A OF CHILD_A OF lunu AND klêna” lines, we would have to give CHILD a special, more general meaning in exactly those two cases only. It is more economical to treat them as copying errors within an otherwise tightly structured family model.

So all of this means the final tree looks like this:

This leaves us with only one term unclear: ungendered “xzor” and gendered “xzorôv”. This could be “person” and “man” or “woman”, but that’s really complete speculation.

The following two sentences also remain untranslated, at the very start of Document 1 and at the very end of Document 4:

• ka BE séš.
• tnav BE ma OF séš.

It is tempting to read this as “I am séš. This is the family of séš.” Tempting as it may be, this is where I draw the line.