Saturday, August 1, 2020

The individual vowels - qamats

This post continues my exploration of the transformations of individual vowels from a pointed text to an unpointed one, particularly to left-to-right SimHebrew where several vowels are made explicit. In the maleh text, vav might sound as u, o, or v. In SimHebrew these sounds are explicit.

In this post I deal with the last step with the qamats, the vowel that becomes 'o' in some cases.

I wondered if it would be better to abandon the very explicit positive framework I have in my code, in favour of a model that converts everything and looks for exceptions. I.e. move from - replace this qamats with an o if the stem and its affixes are in this set, to - replace all qamats with 'o' except for this list. It wasn't hard to make the decision. When I tried removing my positive tests, over 10,000 words in my test data (>25%) and over 1000 distinct stems (>50%) became errors. So qamats is ruled by positive decision. I have found that there is no need to identify qamats qatan specifically since they both transcribe to 'o' anyway.

The process for qamats is in several subsections. We saw in the prior post in this series that qamats plays a role in the double vv and in 21 of the common suffixes. In addition to that, the prior post lists 14 combinations of vowels with qamats that result in double i and one that results in iv. You can also see there the story of qamats with the suffix combination qamats||h (qamats followed by h), which simply sometimes disappears in the transformation, like a puff of smoke in a magic trick. Qamats also interacts with hireq. I will deal with it in a later post under hireq.

Several specific rules apply to qamats. I deal with this as the last of the steps before final touch-up of words.
  1. A couple of stems (pti and yvp) for suffix aim, have the qamats||a dropped.
  2. For alm, ikd, wlv, qamats||v becomes 'iv'.
  3. For hih, hlc, mxa, qlh, qra, and wvb, the suffix t||qamats or n||qamats becomes th or nh. Each stem is restricted to specific affix combinations containing a /t/ or /n/ in the suffix.
  4. For 'tby, clm, pqd, ryl, wlc, initial h||qamats becomes hu (must be hophal).
  5. For akr, m||qamats becomes mo, for ktl, mlk, and prefix vh, h||wamats becomes ho, for ymd with prefix m, the first qamats becomes o.
  6. A general rule follows for strong verbs: qamats in a syllable closed by schwa, or hatef patah becomes o. This rule does not apply to these suffixes: 'u', 'n', 'nm', 'no', 'nu', 'h', 't', 'th', 'ti', 'tm', 'tv', 'to', 'vt', 'vn', 'wyti'. And it does not apply to
    1. verbs with i or v in the second stem position, 
    2. verbs with i in the first position of the stem, and 
    3. ahl, anknv, bkn, bra, brc, crm, dbr, gal, hdr, lbb. 
  7. Finally for open syllables, most of which are limited to specific affix combinations,
    1. 45 stems may have a qamats become an o after the first stem consonant,
    2. and another 25 may have a qamats become an o after the second stem letter.
All of this just for qamats! 

I have in my test data about 1/8th of the Bible covering about 40% of the distinct stems in the Bible. This program will not be finished until the whole text is finished. The mechanics of language are limited and variation even in a closed canon is extensive.


No comments:

Post a Comment