Thursday 8 October 2020

Naming things

 I suppose you have noticed how we all name things, and concepts, and rules, and precedents, and so on. The pattern is old - reaching back to Adam, that earthling, or groundling that eventually gets a name.

When programming and studying a language, I wonder what constitutes a nameable rule. If every rule is specific to a single stem in a single word form, then there would be not much of a system to the language. I am trying to perceive these rules from the back side of the words. I probably have them somewhat wrong. You may remember that I started with the brief summary on Wikipedia. I had developed some initial named rules as noted here in April. They have both reduced and expanded their scope at this roughly half-way point in the game.

I see the Wiki article has been updated this year. If you read around the 'talk' section, you can see there are some shrewd linguistic minds behind the article.

It would be interesting - perhaps when I am finished - to list the exceptions I have found to each of the rules in that article. E.g.

  • Every letter that appears in vowelled text also appears in unvowelled text. [there are a few exceptions, I'm guessing fewer than 100.]
  • After a letter vowelled with a kubuts (the vowel /u/), the letter ו‎ appears: קופסה‎‎, הופל‎‎, כולם‎ [I have 12 stems (+la, cla, gal, hsnah, kla, nwa, qra, rah, rpa, wbval, wvah), 36 instances, drop the qubuts when it precedes an aleph. And there is the occasional other, e.g. the qubuts in ymh in the form l/ym\vt Ezekiel 45:7(23), לְעֻמּוֹת֙.]
  • After a letter vowelled with a holam haser (the vowel /o/) the letter ו‎ appears: בוקר‎‎, ישמור‎. [I have many exceptions to this rule, over 45 stems affecting about 10,000 instances. This may still be a small % of the whole, but it is a significant one.]
  • and so on... 
I have already pointed out the nearly 500 stems (12.5%) that are exceptions to the default hireq rules. And I think there are some rules there that I have not implemented as such but they may emerge. I could eventually compare the two versions of the corpus with dominant rules and exceptions by percentage. It would take some imagination if I have any left after finishing the task at hand.

My overall somewhat vague question is: What is the set of abstractions by which we can have the optimum description of the language?

Perhaps the question is relevant. I would still note that the missing vowels in the unpointed text are critical to singing the words. 

(Non-Hebrew singers must have a text that allows them to know the vowels to sing. Why would anyone sing the text who does not know Hebrew? For the same reason that one would sing Bach in German even if one does not know German. The music fits the tongue it was written for better than it does a translation. And don't say we have no original music - it is in the text even if we disagree on how to read it. And it can be added to the unpointed text.)

No comments:

Post a Comment