Hebrew rules of shortening could be
generally formulated thus. Only kamatz and tzere regularly turn into schwa, and
also possibly holam (the vowels based on the a sound).
Stressed suffixes shorten one of the above
vowels and only them to schwa. Vowels are shortened right-to-left. Two
suffixes, such as plural and possessive, shorten two vowels. Smihut is a
specific case of this rule. The second word of smihut, stressed syntactically,
acts similarly to stressed suffix, shortening one vowel to schwa. This is how
clavim became clvei-calvei in smihut.
Unstressed suffixes shorten the nearest (leftmost)
vowel. This is a mnemonic rule. Unstressed suffix closes the preceding syllable
which causes the shortening, diber (final consonant does not firmly close)
Unvowelized suffixes do not shorten. However,
the resulting CVCC syllable, even if stressed, shortens the long vowel to patah.
Note that for cotevt it was chosen to destroy the super-heavy syllable by
turning it into cotevet, instead of shortening tzere into patah, as in the 2fs
Vowelized prefixes (similarly to unstressed
suffixes) shorten the nearest vowel to patah, like in hitpael.
Unvowelized prefixes (similarly to unvowelized
suffixes) should have had no effect, but for phonetic reasons, CCVCVC, turned
into more equally weighted CVCCVC, similarly to loan word drachmas turning into
the Hebrew darkemonim, tcatov - tictov. Silent schwa in the prefixes is also
indicated by dagesh after the first CVC syllable. When the schwa is mobile, as
in the plural smihut form divrei (dvrei), there is no dagesh after CVC.
But unvowelized prefixes produced metathesis, with first vowel moving to the
prefix, and silent schwa to the next consonant, thus producing dagesh in the
second root consonant because of the silent schwa; long vowel became short hirek
in the closed unstressed syllable.
In several specific cases, holam does not
shorten, notably in the nouns of shalom form. This case is easily explained by
incorrect adding of suffixes. The words of shalom class are actually
adjectives-turned-nouns. The only appropriate suffixes for them, like for other
adjectives, are gender or quantity monosyllable suffixes. As such, they shorten
only the first syllable. This grammatical tradition was preserved when these
words acquired two-syllable suffixes unique to nouns, such as caim cem, and
only the first syllable shortened, shlomchem (not schlamchem like dvarchem).
To use the rules of shortening, we need to
define the rules of accentuation. They are simple, with few exceptions
explainable by grammatical deviations.
The Hebrew accent is always on the last syllable
(containing the CV segment), but on tzere (or ex-tzere segol or patah) if it is
available. Suffixes of V or VC type override the tzere rule.
The V-VC rule occurs to avoid open stressed
syllables: CvCEC.v would have turned into Cv.CE.Cv. This should have brought
about the introduction of iud to close the stressed syllable with tzere in
talmidEnu -> talmidEinu fashion. For some reason, this solution was unwelcome,
such as in 3fs verbs.
One would expect holam, the au diphthong, to
draw the accent the same way as tzere, the ae diphthong, does. This, indeed,
happens, except in several irregularly constructed words.
The nasal consonant nun introduces a bit of
variance. Since CNV syllables are phonetically odd, segol is introduced between
the consonants, and lengthened in tzere when accented. No iod after the accented
tzere because it is actually segol.
The waw-reversing shifts the accent to conform
the words to iambic rhyme, and also perhaps to distinguish them from
non-reversing conjunction waw. All Hebrew words, theoretically, before
transformations, adhered to iambic stress.
An unusual stress in hifil verbs is due to the
obscured tzere in prefix, resulting from the a sound in hey, and its vowel hirek,
airgish Ergesh - irgIsha.