language and script 103
/yaʿʿol/ ‘he enters’ (Kai 222 B: 35, “imperfect”). dstem forms inflect
like sound roots; perhaps the same applies to the gt and dt stems, but
evidence is lacking.
4) Verbs with a long vowel between the initial and the final root conso
nant (“hollow roots”) preserve this vowel in the gstem “imperfect,”
otherwise the corresponding long vowel of the sound verb appears: qm
/qām/ ‘he rose’ (Kai 202 a: 3, from qūm) but ymwt /yamūt/ ‘he dies’
(Kai 224: 16, from mūt). presumably, this vowel was shortened in the
final syllable of the “short imperfect,” as the difference between lšm
/laśim/ ‘may he erect’ (Kai 309: 11, from śīm) and yšym /yaśīm/ ‘he will
erect’ (Kai 309: 12) in tell fekheriye suggests (see the corresponding
remark in the section on verbal conjugations). however, later vocal
izations do not indicate that the vowel of the “perfect” became short
before consonantal afformatives, in contradistinction to canaanite
and classical arabic, hence a form like wrṣt ‘and i ran’ (Kai 216: 8,
from rūθ ̣) presumably has to be vocalized /warāθ̣t/. the gstem active
participle and the entire dstem of most verbs behave like sound roots
in later aramaic, but the situation cannot be assessed for the earli
est attested stages. it is not impossible that some verbs replaced the
dstem by another pattern based on reduplication of the final root
consonant (/qawmem/ in the “perfect” for qūm).102
5) Verbs with a rootfinal /ī/ seem to preserve this long vowel in all “per
fect” and imperative forms (perhaps shifting it to /ay/ with /ī/ and to
/aw/ with /ū/ of the afformatives, as in later aramaic varieties).103 in
the “long imperfect,” the participle, and the gstem infinitive, how
ever, wordfinal /ī/ changes into /ɛ̄/, whereas the “short imperfect”
has /ay/, hence the distinction between thwy /tahway/ ‘may she be’
(Kai 222 a: 25, from hwī ) and yhwh /yahwɛ̄/ (< /yahwī/) ‘he will
be’ (Kai 223 a: 4), later lost in aramaic (see the discussion above).104
102 the “perfect” knn ‘he set up’ from kūn in tell fekheriye (Kai 309: 10) and the corre
sponding “short imperfect” in the following line, however, could also be parsed as dstem
forms of a variant geminate root knn (cf. Beyer 2004: 332, assuming that the socalled
“lengthening stem” with the expected “perfect” /kānen/ for the root kūn emerged, but in
much later stages of aramaic).
103 Beyer 1984: 489, but cf. nebe 2010: 319 on the spelling qnt ‘i have acquired’ in the
Kuttamuwa inscription (l. 1), which seems to point to /qanīt/ rather than /qanayt/, since
the latter would normally have been written qnyt.
104 a form yhy of unclear significance occurs in the Kuttamuwa inscription (l. 7; see
pardee 2009a: 68). the syntactic environment (protasis of a conditional construction) sug
gests a “long imperfect” (nebe 2010: 325, 329–330), in which case this would be a defective
spelling of a root hyī ‘to be’, otherwise unattested in aramaic and Samʾalian (see section 6,