G. Emphatic Sounds
11.9. There are some examples of inserted b or p after m, which can
not be interpreted as dissimilation of gemination, e.g. Palaeosyrian zumūbaru, broken plural (§31.28) of *zambāru < zamāru, "song";
Gogot tambuyà next to tamuyà, "orphan"; Amharic qâmbár from Ge'ez
LABIALS 113
qamar, "yoke"; Spanish Alhambra from Andalusian Arabic 'al-Hamrā', "the Red (Castle)"; Apppt transcribing Hebrew 'mry,
"Omri"; IappXaxos used for Yamlik; Zap\|/ai marking the proper name Šamšay; Mandaic 'mbr' from 'mr', "sheep". A similar phenomenon is attested in Cushitic (e.g. Boni šimir, Somali šimbir-ta, Rendille cimbir,
"bird") and in Indo-European languages, as in French chambre from Latin camera or in Greek au^poxos, "immortal", from popxos, "mor
tal". Therefore we cannot be sure that Greek X,áiiJ35a reproduces a Semitic pronunciation of lamed. The original nasal may disappear in front of the inserted plosive, as in the secondary Greek form ppoxos,
"mortal". It is quite probable therefore that Eblaite si-pis and Ugaritic špš, "sun", originated from *śampšu or *śimpšu, with an inserted p after m, like in Ect|ii|/ai. However, also a secondary m may be inserted before b, as in East Gurage dmbab, "snake", and in the name of Humbaba, both related to Harari and Arabic hubāb, "snake", and both attested also with the change b > w (Argobba hzwaw, "snake"; Old Babylonian Huwawa). But no evident case of such an insertion seems to occur before p.
11.10. The spirantization of labials, dentals, and velars in the various Semitic languages is well known. Although Aramaic p and k are never rendered in Demotic by / and h, the spirantization is probably indicated by the Greek transcriptions of k, p, t in the Bible: the Septuagint tran
scribes these letters either by K, TT, T, or by %, 9, 0, but these two series of transcriptions do not imply regular positional variants as in the tradi
tional Jewish reading of Biblical Hebrew and Aramaic, where the non-geminated consonants b g d k p t are spirantized in post-vocalic position.
These positional variants are attested in the Middle Ages in all Jewish Arabic-speaking communities, including Spain, and in France. With the exception of a few communities, only some of the six consonants are realized nowadays in Hebrew as plosives and as spirantized or labioden
tal fricatives. Among the Samaritans, b g d kp t have survived only with one pronunciation except for p, which is pronounced b when geminated and mainly as/elsewhere. Instead, traces of the double pronunciation of b g d k p t can be detected in Neo-Aramaic, both Eastern and Western, but the plosive and the spirantized realizations have both attained phone
mic status and are no more conditioned by their position. In Tūroyo, however, there are hesitations in the pronunciation of p / / . In Eastern Syriac, p is never spirantized, except in the loanwords of some dialects.
Arabic does not know spirantization of labials or velars, but the labials
b, m, and the velars k, q can be spirantized in Ethiopic when they are not geminated. The spirant consonants are not phonemic in Ethiopian lan
guages and they can appear as free variants. In Śheri (Modern South Arabian), neither b nor m may occur in intervocalic position, but there is either a compensatory lengthening of the vowel or a raising of a semi
vowel, facts that seem to imply a previous spirantization of the labials.
In Assyro-Babylonian, spirantization of labials and of dentals cannot be detected, except in the case of m (§11.8), but there are some sporadic traces of spirantization of non-geminated velars k, g, indicated by signs with h; they are not necessarily connected with post-vocalic position (e.g. the divine name Nusku = Nušhu).
1 1 . 1 1 . The labial semivowel w has regular correspondences in all the Semitic languages. It may come from non-geminated b or m by spiranti
zation, and from rounded phonemes bw, gw, kw, hw, qw in Ethiopian lan
guages (e.g. Gurage wáz < bwàz, "slave"; cf. §18.7); it may also result from a secondary diphthongization of a long vowel. The labial w can serve as a glide between vowels, especially after u (e.g. Assyro-Baby
lonian pa-nu-ú-a = panūwa - panū'a; Arabic maqrū'a > maqrūwa,
"being read"), and also as an on-glide in initial position before o, u (e.g.
Ethiopic wof and of < 'op, "bird"), and even before a (e.g. colloquial Arabic wakkil < 'akkil, "he fed"; cf. §19.24). In various diphthongs w may be reduced to a vowel, generally 6 or ū, but also ā (aw > ā) (§ 22.1,3-4; 27.23-24). Similarly, Ethiopic labialized consonants fol
lowed by à may alternate with consonants plus -o (<-o); e.g. Amharic Gwàġġam or Goġġam, the name of one Ethiopian province.
1 1 . 1 2 . Phonetic w occurs as a speech-sound throughout the life of East Semitic, but its graphic notation in cuneiform syllabic scripts is imper
fect and doubts have been cast on its phonemic status from the Old Babylonian period on. Since the phoneme w did not exist in Sumerian or Pre-Sumerian, cuneiform writing does not have any special signs to express it. The Semites were forced therefore to find ways of expressing w in their writing and they regularly used the sign PI in the function of wi, wa, wu (e.g. sá-pu-wa-an, "flexible [shoes]"). The initial wa was sometimes expressed also by u+a or á+a, as in Ú-ar-ti-a /Wardiya/, and by the sign É having the value 'à, as in 'à-ba-al /wabāl/, "transporting".
Replacement signs with b and m were also used, as in Old Assyrian Tan-bar-ta and Tan-mar-ta, which are allographs of Tan-wa(Pl)-ar-ta.
These replacements show that the scribes were aware of the phonetic
LABIALS 115
correlation of the labials b, m, w and that the apparent changes w > m and w > b in East Semitic are to be considered as graphic replacements or allographs and not as real phonetic developments. The sign PI having become restricted in later periods to the values pi, pe, except in archaiz
ing script and in peripheral regions, its place in the system was taken by signs with m and b, probably following the occasional spirantization of these phonemes in the spoken language. In Middle Assyrian and in NeoAssyrian, also the vowelsign u could be used to indicate wa, as in M -nu-u for Ni-nu-wa, the other possibility consisting in not expressing w at all, as in Ni-nu-a, "Nineveh". The loss of w at the beginning of words can generally be assumed from the Old Babylonian period on.
11.13. A development w > y in initial position characterizes Amorite, Ugaritic, Aramaic, as well the "Canaanite" of the second and first millennia B.C. E.g. yld, "to bear", is in opposition to wld in the other Semitic languages. The use of the cuneiform sign PI to mark yi, yas, yú besides wi, wa, wu (§11.13) in texts influenced by Amorite, Ugaritic, and Canaanite (Amarna correspondence) witness to this development which had a repercussion on scribal habits. Initial w is preserved only in the conjunction wa-, "and", in the name of the letter wāw, meaning
"hook", in a few loanwords and in foreign proper names. A South Ara
bian Sabaic inscription shows a dialectal tendency to replace w by y at the beginning of words and the same phenomenon is attested once in an Arabic Hidjazi poem where yāzi'ahum, "theif commander", stands for wāzi'ahum. In Sabaic, fluctuation between the semivowels w and y is sometimes seen also in medial and final positions, e.g. in kyn against normal kwn, "to be'\ in rdw and rdy, "goodwill". The same fluctuation occurs also in Andalusian Arabic, e.g.fawha and fayha, "fragrant ema
nation", hawba instead of usual hayba, "gravity". It is encountered in medial position after a consonant in the Hebrew and Aramaic word
'aryē, "lion", instead of common Semitic *'arwiy-.
11.14. In Arabic there is a possible development wu- > u- and w > i-in i-initial position, especially i-in Hudhail, a Hidjazi dialect. These forms are written 'u- and 7- in the Arabic sources, but the hamza is there a purely orthographic feature. Besides, some roots with first radical w have sporadically a variant with ', e.g. 'irtun, "inheritance", from warita, "to inherit" (cf. §19.24). There is also the regular Arabic prac
tice of substituting ' for w/y after ā, e.g. *ġāwiz > ġā'iz, "lawful",
*'iġrāy > 'iġrā', "enforcement".
11.15. The labiodental/may result in certain conditions from the inter
dental t and from a lisping articulation of š I s so that the sound produced is [0] > [f] (cf. Greek Oepp.os, Latin formus, "hot"). This phenomenon is well-known to Arab grammarians and enters in their category of 'ibdāl luġawī or "lexical substitution"; e.g. ġadat > ġadaf, "grave, tomb";
Masyaf for Mediaeval Arabic Masyat; Greek transcription <í>épe7i of the Syrian place name Tārib; Lihyānite Rubaf for Rubat. This phenomenon would explain the Egyptian pronominal suffix -/ of the third person mas
culine singular and the Argobba prepositions wdfc, "inside", when com
pared with Harari ustu, tef, "under", which possibly goes back to a *tes alternating e.g. with Amharic tacc, and then by analogy lef, "on".
3 . D E N T A L P L O S I V E S
12.1. Common Semitic or Proto-Semitic has two dental plosives,