checksumerian

The Sumerian language is rather interesting. It has a high number of homophones (as deduced by co-translations with Akkadian), and one of the suspicions is that it was at least partially a tonal language like Chinese.

Another interesting feature of Sumerian is that it is an ergative-absolutive languages. That is to say that the subject of an intransitive verb (i.e., "Bob sleeps") and the object of a transitive verb ("Bob kisses Jane") are treated the same, as they are not the agent of an action. subject of a transitive verb ("Bob kisses Jane") is marked with the ergative case. This is different from most languages, called nominative-accusative language, where Bob is treated the same because he is the subject of both sentences.

But the thing about Sumerian that I find the most interesting is that it appears to have a checksum at the end of each sentence. Here is generally how the language is structured:

[Noun Phrase].[case marker] ... [Noun Phrase].[case marker] ... (etc) ... [Verb Phrase]

A Noun Phrase is a noun and its qualifiers (adjectives, possessives, etc.) For example, "my green house" is a noun phrase. Cases in general can be dative ("for X"), ablative ("from X"), locative ("at X"), terminative ("to X"), ergative ("agent of X") etc.

What is the interesting part is the verb phrase. The verb phrase consists of the following particles, in this order:

- a term that indicates the general mood of the sentence (indicative, subjunctive, optative)
- another term that people aren't entirely sure about. One theory is that it has something to do with deixis (near/far in space, near/far in time, near/far in social stature, etc.) It isn't translated, usually, because no one knows what it exactly means.
- Another set of terms that is a cross-reference the previous cases. It is wholly redundant and always occurs in a certain order (e.g., dative first). It essentially serves as a checksum for the previous noun phrases.
- Another term that indicates the relationship of the agent (causer of action) to the patient (the recipient of the action): animate/inanimate, 1st/2nd/3rd person, etc.
- The verb.

This checksum interests me, because I have a personal hunch that it be related to the number of homophones in the language. For example, "e" means "Palace" and "e" also is the marker of the ergative case. So the phrase lugal-e e mu-un-du means "The king built the palace". (mu-un-du is the verb phrase.)

The two "e" in a row can be problematic without a checksum at the end ("the "un"). It could be a tonal difference, or it could be resolved with a checksum to say "yes, there was a palace, and it also was the recipient of an action of being built."

I wonder if any other languages have similar checksums.