My point in my last posting was that qosht., nerd, and 'ard show
exactly the same behavior as h.es., h.oq, and `am -- qosht. has h.olem
and so does h.oq; nerd has s.ere, and so does h.es.; 'ard has
pathah. in "context" and qames. when pausal, and so does `am. So a
compensatory lengthening explanation wouldn't explain the parallel
vowels in a parallel way.
Syllables which would have main-stress, except that the word in which
they occur has been joined with maqqeph to the following word, differ
both from surface main-stressed syllables, and from syllables which
never would receive main stress. In a true non-word-final unstressed
syllable, an underlying non-low front short vowel would generally
become [i] in a closed syllable, while the same doesn't happen in
syllables which would receive main stress except for maqqeph.