Here we look at hireq, when it is ignored and when it appears. There was an overall strategy. I had several rules based on prefix to suppress hireq. They almost all disappeared when I figured out the closed syllable rule.
- Suppress hireq
- when there is an /h/ in the prefix and the stem is in hlc, hll, clm, ixg, ixt, lvi, npl, pla, rah, rgy, wlc, wmy, wgh, wqh, xmt, yll.
- or the stem is in hlc, irw, nwa, rbb with conditions for some affixes.
- Suppress hireq when it is in a closed syllable,
- if the last character of the prefix is /n/
- and stem is one of ptk, ckd, clm, csh, dmh, mlt', pla, pzr, rah, rpa, wbr, wck, wmy, and the first syllable of the word is closed with a schwa.
- or any stem with a weak first character and the syllable is closed with a hatef-patah (only one instance of this).
- if the stem is not in /n/ and the stem is not in bqw, clm, csa, gll, irw, igy, lbb, nwa, rxp, qll, qvh (or it is qvh with the last character of the prefix a /t/).
- There is an exception for words from stems hlc, wal with a final schwa.
- Allow hireq to be come 'i*i' - (which will later translate to double i).
- for words beginning with v||schwa||i||hireq where the stem begins with i and the word is not the name of a person, location, or people (I do have this explicit semantic information in my data),
- for some stems formed with weak consonants (ibw, iry, ixr, n'ty, ndr, lvn, ain, ild, nwa) with conditions on the affixes,
- for the long list of stems (see this post) that allow i with prefixes i, vi, iv, hiv (happens with stem idy),
- Allow hireq to become i
- for several lists of stems with conditions on the affixes.
- for stems with yod as part of the stem, excluding any stems, another long list, where hireq is explicitly prevented.
- Much later in the process, the last steps,
- prevent double i where the stem does not allow it.
- prevent the weak hireq from forming a syllable with a common single or double prefix unless this is overridden for a stem.
Though the prevention logic has simplified considerably, this whole process seems too complex still.
No comments:
Post a Comment