Invitational round | 15 points | 13.00% | Problem statement | Official solution | Tags: Phrase translation
Group the examples by structure:
Obviously, pronominal possessives are absorbed into the noun, while "nte" is a genitive marker. We don't know yet if the possessor is after "nte" or before it.
Compare "two chickpeas" and "a chickpea of the rich man"; the part that repeats, "gafkaf", means "chickpea". Therefore the prefix "ou-" means "one/a"; "snau" means "two"; "permmao" means "rich man". This tells us that the possessor is after "nte".
In "thome mpefson snau", "thome" and "mpefson" mean "cup" and "brothers" in some order. Given that "paftoou nson" = "my four brothers", it seems most likely that "-son" means "brothers" as the only part that repeats. However didn't we just decide that the number "snau" goes after the head noun? This isn't unique though—"maab nanaš" = "thirty oaths" and "pesanaš" = "her oath" also only has "-anaš" as the repeating part, further suggesting that the number goes before the head noun for cases other than 2. So we have no better choice but to say that "snau" is different and is the only number that comes after the head noun. Does "paftoou" stand for "my four" or just "four"? I go with the former because "nson" looks too simple while "paftoou" looks more complex. More precisely, I hypothesize that the pronominal possessive affix always happens on the first word of the noun phrase, whether that's the number (in case of > 2) or the head noun (in case of 2).
In "oucij nte oushime nte praše", we already know that "oucij" is the possessed "a hand". The rest, "oushime nte praše", must be "a joyful woman". "ou-" means "a"; "shime" appears again in "teshime" = "the woman", so it has to mean "woman", leaving "nte praše" = "joyful" ("of the joy").
Now we know the structure of genitive noun phrases, we can say more about "nte". If one glances at the examples with "nte" vs. those without, note how the ones without all start with "the", while the ones with "nte" all start with "a" or "some". This suggests that "nte" is only used if the possessed is indefinite. On the other hand, the no-"nte" examples all have the possessive start with a nasal, while none of the "nte" examples do. This nasal is very marked in this context because in almost no other case do we have such a phonologically unnatural sequence. This suggests that the nasal prefix is the equivalent of "nte" in the no-"nte" examples, and the nasal prefix becomes "m-" if before another labial consonant ("p" in "mpekhto", "mpefson", "mpjahj") and "n-" otherwise.
So at the phrase level, here's the syntax:
The [determiner] includes pronominal possessives and articles. Let's remove these genitive and number words to analyze the bare noun phrases. First consider those nouns that appeared multiple times:
For "horse" and "brother", there seems to be some spelling variation that I'm not sure how to account for: "htōōr" vs. "hto", and "snēu" vs. "son". Despite this, it's clear that there's a prefix that stands for the determiner (possessive, quantity, etc.) and the root. As aforementioned, the "ou-" prefix is the indefinite singular article "a". Similarly, the "hen-" prefix is the indefinite plural article "some". Here are the remaining nouns:
Notice how the remaining words all start with one of "n", "p", or "t". The "n-" prefix is used for all plural cases, including in number constructions ("thirty oaths") and possessive constructions ("your brothers"). The "t-" prefix and "p-" prefix are both used for singular nouns, but "t-" is used for feminine nouns: woman, mother, head, cup; "p-" is used for masculine nouns: man, father, brother, horse, oath, gnashing, place, joy. In number constructions, the number gets the gender marker, while the noun gets the plural marker.
Remove these definiteness/gender/number markers and list them again.
Therefore, the next part encodes the possessor: my = "af-", your (sg) = "ek-", his = "ef-", our = "en-", your (pl) = "etn-", their = "eu-". There are two cases with a lone "e-" prefix that doesn't really stand for anything, so I suppose that "p-"/"t-" and "pe-"/te-" are allomorphs, where the "e" is added for phonotactics.
So to recap, the noun morphology is as follows:
[definite/gender/number][possessor][root]
a. p-ef-toou n-time = Masc.His.four PL.village = "his four villages"
b. n-gafkaf n-n-en-eiote = PL.chickpea Gen.PL.our.father = "the chickpeas of our fathers"
c. hen-anauš nte p-a-son = some.oath of Masc.my.brother = "some oaths of my brother"1
For L2:
a. your(sg) thirty horses = Masc.your.thirty PL.horse = p-ek-maab n-hto2
b. the horse of my joyful brothers = Masc.horse Gen.PL.my.brother Gen.Masc.joy = pe-hto n-n-af-son m-p-raše
c. a chickpea of your(sg) mother = a.chickpea of Fem.your.mother = ou-gafkaf nte t-ek-maau
d. two horses = hto snau
According to this analysis, the "my" morpheme is actually "-a-" and not "-af-", so "four" is "ftoou" instead of "toou". With this analysis, (a) is actually "p-e-ftoou n-time" = Masc.four PL.village = "the four villages". If we consider the problems to be part of the input data as well, then this analysis makes more sense. Then in L2(b), we should also write "n-n-a-son" instead of "n-n-af-son". ↩
I'm unsure about all the spelling variations such as whether it should be n-hto or n-htōōr here. ↩