A proper view of Arabic, Semitic, and more.

• Author: Rendsburg, Gary A.

The recent contribution by George Mendenhall in the pages of this journal (1) is so riddled with errors and idiosyncratic views, it demands a response. The author's main contention is that the roots of Arabic are to be found in the language(s) of the Late Bronze Age Levant (Ugaritic especially). In his own words: "It is here suggested, accordingly, that the origins of the linguistic phenomena characteristic of Arabic are to be located in the population of Syro-Palestinian groups who, in response to the increasing turmoil and violence of the Late Bronze Age, migrated south to the relatively remote and untouched regions of Arabia. Thus, instead of viewing Arabia as the early homeland from which the later Semitic language groups departed, we should view it as a late refuge to which population groups from Syria and Palestine migrated. They took with them, of course, their material culture, and above all their Bronze Age linguistic repertoire" (p. 18).

Now, to be fair, there is nothing impossible in the scenario that Mendenhall presents--except for a) the lack of any archaeological data to support this view, and b) the specious and spurious nature of the linguistic arguments that he musters. Let us examine it in detail.

1. Mendenhall relies heavily on his reading of the Byblos Syllabic texts, (2) but he has convinced few scholars of his "decipherment." In the words of one critical review, "Mendenhall's imagination is vast enough to enchant us with etymological possibilities. Unfortunately, this is not enough to convince us as to the validity and success of his decipherment. Far-fetched etymologies, forcing of known and accepted linguistic rules of other well known languages, together with much too much unreliable data, play a major role in this scholarly work. This makes Mendenhall's decipherment of the enigmatic inscriptions from Byblos no less enigmatic than the inscriptions themselves. I am afraid that until a bilingual or another 'tripod' appears on stage, there can be no proof for this (or any other) deciphering attempt for these inscriptions." (3)

2. Mendenhall claims that diglossia "is attested not only at Late Bronze Ugarit, but has also become increasingly in evidence from pre-Islamic Arabic inscriptions" (p. 18). It is hard to imagine what evidence he would muster to defend this statement. Both corpora, especially the latter, are extremely limited. Now, many linguists would argue that every language in the world with a written tradition is characterized by diglossia, since people always write and speak differently. So it is very likely that diglossia existed in ancient Ugarit and in pre-Islamic Arabic, (4) but the data are lacking to demonstrate the point and to highlight the lexical and grammatical differentiations that presumably existed between the literary and colloquial registers in these two instances. Contrast the effort (by one of the present authors) to demonstrate diglossia in ancient Hebrew, with a much larger corpus and with sufficient data to (hopefully) have proven the point. (5)

3. A statement such as "Ostracon 2 from Kamid el-Loz is Arabic" (p. 20) is without justification. The ten inscribed sherds found at Kamid el-Loz are written in an alphabet ("altkanaanaischer Schrift" according to Gunter Mansfeld, who first published the texts) with some signs that resemble their ancient South Arabian equivalents, (6) but that does not mean that the language of this inscription is Arabic--unless, of course, one accedes to Mendenhall's view that anything Arabian is by definition Arabic. More importantly, Ostracon 2 has only five signs, and of these only three appear to be actual letters. Mansfeld read a single word here, as either btm or rtm or dtm, with the following comment, "wenn auch keiner dieser Stamme in den altsemitischen Sprachen bisher bekannt ist." (7) True, Mendenhall reads the inscription in the opposite direction, with the five symbols yielding lmtry (p. 20), but in no way can this term be deemed to be Arabic. And even if it were, no linguist would weave such a major hypothesis (of Arabic origins) from a single thread (an ostracon consisting of three to five graphemes). (8)

4. Mendenhall denies that gayin is a proto-Semitic consonant. He states, "As a matter of fact, there is no epigraphic evidence whatever for ghayn in Semitic until the LB texts of Ugarit, and it exists nowhere but in Ugaritic and the proto-Arabic language complex" (p. 21). This statement is patently wrong. But before addressing the issue itself, we first note that Mendenhall seems totally unaware of the history of research on this matter. (9) As is well known among Semitists, already more than half a century ago, Rudolf Ruzicka proposed that /g/ is not a proto-Semitic consonant, but rather was an Arabic innovation. (10) Ruzicka's position, in turn, gained further support from Karl Petracek. (11) The response to this view came from Otto Rossler, who, in two classic articles, demonstrated that /g/ is indeed a proto-Semitic consonant. (12) Astoundingly, nowhere does Mendenhall refer to these earlier studies. (13)

   And now to the issue itself. Recent research (building on Rossler's earlier studies) has established beyond a reasonable doubt that /g/ was a separate phoneme in early Akkadian, distinct from both /'/ and /h/. In Old Akkadian the recently borrowed writing system was not equipped to represent /g/ consistently, but clear examples are written with the HA/HI/HU series of signs, e.g., za-ha-ar-tim /sagartim/ 'small' (feminine singular adjective), za-ah-ra /sagra/ 'they (dual) are small,' both from the root sgr. (14) In Old Babylonian, writings of the reflex of Proto-Semitic /g/ vacillate between the H-signs and [empty set]; for an example of the former, see Akkadian halapu /galapu/ 'cover, close,' cognate with Arabic gallafa 'wrap, cover,' Ugaritic glp 'husk' (cf. Hebrew [TEXT NOT REPRODUCIBLE IN ASCII] 'cover, wrap'). (15) As this illustration demonstrates (unlike with the reflex of /'/), no vowel raising occurs in the vicinity of /g/, thus providing further evidence that the two consonants were originally distinct.

   Also relevant are the data from Amorite, with the following personal names attested: ha-za-la and a-za-lu-um = /gazal-/ 'gazelle'; hu-za-lum and u-za-lum = /guzalum/ 'little gazelle'; and pu-ur-hu-sa-nu = /purgusanu/ 'flea'--yielding the conclusion that "/g/ [durfte] im Amurritischen als selbstandiges Phonem erhalten sein." (16)

   When we turn to the world of greater Canaan, we note that Egyptian transcriptions of Semitic words demonstrate beyond doubt that "/g/ is a phoneme distinct from 'ayin, as it is consistently transcribed by Egyptian q and g in roots containing Proto-Semitic */g/. ... The evidence clearly indicates that /'/ and /g/ were phonemically opposed in most [Canaanite] dialects." (17) Further evidence for this distinction derives from Greek transcriptions of Hebrew words in the Septuagint. In the words of Joshua Blau, who thoroughly researched the topic, "G [= the Greek transcription] reflects a language in which Proto-Semitic ' and g were still separate phonemes." (18) That is to say, as late as the third century B.C.E. (the date of the Septuagint), Hebrew continued to distinguish the two phonemes, even though only a single grapheme [TEXT NOT REPRODUCIBLE IN ASCII] existed to express the two consonants in writing. Apparently, Mendenhall is totally unaware of this evidence, given his statement (cited above) that /g/ "exists nowhere but in Ugaritic and the proto-Arabic language complex." (19)

   Next there is the evidence of Old South Arabian, whose alphabet includes a separate grapheme for gayin, though apparently Mendenhall includes this language in his "proto-Arabic language complex." Given the many important grammatical differences between Old...