language and script 83
/rāθ̣t/ ‘i ran’ (Kai 216: 8) sounded like [rātt].30 there is also some evidence
for the dissimilation of the first of two emphatics in the same word, the
best example being kyṣʾ ‘summer’ (Kai 216: 19) instead of the expected
qyṣʾ from the original /qayθ̣/ (written qyṭ in later aramaic). it mostly
seems to target /q/ > /k/, but not consistently (cf. llqṭw /lalqoṭū/ ‘let them
collect’ in Kai 309: 22). however, the true extent of this phenomenon
remains elusive, as dissimilated forms appear side by side with regular
ones even in the same text (e.g., ṣdq ‘justice’ in Kai 216: 4–5).31
other sound changes are rather difficult to pinpoint. there is general
agreement that syllablefinal /ʾ/ disappeared at some stage between old
aramaic and official aramaic, thereby causing compensatory lengthening
of the preceding vowel if that vowel was short. as a consequence, (most)
verbs originally ending in /ʾ/ merged with those ending in /ī/. the graph
eme {ʾ} was often preserved in historical spelling, especially for what has
become /ā/ in the emphatic state, and eventually triggered the use of {ʾ}
as a vowel letter for /ā/ even in cases where that vowel did not originate
from the loss of a glottal stop. furthermore, the grapheme {ʾ} for an ety
mological glottal stop could drop out or be replaced by {h} (the normal
means of indicating /ā/) in the emphatic state in less formal orthography.
cuneiform transcriptions of personal names may imply that this process
began in the second half of the 9th century B.c.; direct evidence from
aramaic, however, appears only gradually.32 aphaeresis of wordinitial
/ʾ/ in the numeral ‘one’ ḥd /ḥad/ (< */ʾaḥad/), fem. ḥdh, by contrast, is
common to all aramaic languages from the outset, whereas prosthetic /ʾ/
(especially with sibilants) seems to be merely incidental (cf. ʾšm /ʾešm/ in
30 cf. Brockelmann 101965: 18 (§§ 26–27), hence pšyṭtʾ ‘the simple one (fem.)’ of course
has to be transcribed pšīttā.
31 regressive dissimilation of two emphatics in aramaic resembles geers’s Law in
akkadian. it appears to be more frequent in Mandaic (which was spoken on the territory
of Babylonia and seems to exhibit other traits of Babylonian as well, such as phonetic
degemination) than in other aramaic varieties (gzella 2008: 97 with n. 38). one may thus
entertain the possibility of substrate influence of akkadian pronunciation, but it is impor
tant to note that this dissimilation does not produce identical results in akkadian and
aramaic (Kaufman 1974: 121f, who rejects the hypothesis of a contactinduced feature). See
folmer 1995: 94–101 for likewise sporadic evidence from later periods.
32 Beyer 1984: 104–106 and id. 2004: 52f. the earliest possible example is z /ðā/ (< */ðāʾ/)
‘this one (f.sg.)’ in the pedestal inscription from tell halaf in assyria (Kai 231) instead of
the usual spelling zʾ. no uncontroversial examples of {h} for /ā/ in the emphatic state
are yet attested in the oldest aramaic texts, although this became more common in later
periods; the only possible case, mlkh in the graffito Kai 203 from Syria, can be explained
differently (degen 1969: 8 n. 40; e.g., ‘his king’ instead of ‘the king’).