Leonid Kogan Russian State University for the Humanities Moscow COMMON ORIGIN OF ETHIOPIAN SEMITIC: THE LEXICAL DIMENSION* Introduction Semitists have a tendency to be sceptical with regards to genetic classi- fication and negative statements about «the questionable usefulness of classi- ficatory schematizations» (Renfroe 1992:7) can easily be found on the pages of Semitological treatises. In spite of this trend, various aspects of genetic classification of Semitic have always occupied many of the best minds work- ing in this branch of comparative linguistics. A proper evaluation of the sub- grouping procedure as applied to Semitic languages has been admirably out- lined in a recent survey by John Huehnergard: «Classification and subgrou- ping of language families are among the most important of the comparativist’s tasks, and this obtains in our field, too... Indeed, classification and subgouping should inform comparative work and historical reconstruction, for these activi- ties are inextricably interwined ... In other words, classification is not simply a mind game...» (Huehnergard 2002:130).1 A coherent classificatory pattern of Semitic has two facets: several major splits and unities are to be postulated and proved, such as East Semitic vs. West Semitic or Central Semitic vs. South Semitic; simultaneously, the com- mon origin of each minor subdivision (Cannanite, Aramaic, ESA, Ethiopian, * I am deeply grateful to Maria Bulakh and Alexander Militarev for their critical remarks on a preliminary draft of this article. The work on the present topic was carried out within the project «Studies in the Genetic Classification of Semitic» supported by the Center for Fundamental Research (project No. ÐÄ02-3.17-101) which deserves my most sincere gratitute for its assistance. My work on the present contribution was carried out in the framework of the projects 03-06-80435-a (ÐÔÔÈ) and 04-04-00324a (ÐÃÍÔ). I am grateful to both institutions for their help. 1 In reply to the following statement from ULLENDORFF 1961:30: «Classification is harmless, unobjectionable, and at times even useful if limited to describing present- day habitat and the prevailing geographical circumstances, but it becomes positively dangerous, i. e. obscuring rather than illuminating, if meant to explain genetic con- nections». [For the list of abbreviations used in this article see p. 392.] Ullendorff’s position is rightly qualified by Huehnergard as «without parallel in comparative work in other language families». For another highly positive assessment of linguistic classi- fication v. GOLDENBERG 1998:461, quoting such outstanding figures of general and Indo-European comparative studies as Baudouin de Courtenay, Meillet and Greenberg. Downloaded from Brill.com10/02/2021 09:12:28PM via free access 368 Scrinium I (2005). Varia Aethiopica MSA etc.) is to be demonstrated as clearly as possible. Demonstration of the diachronic unity of Ethiopian Semitic (ES) is a model illustration of the se- cond task: a compact, geographically and historically isolated group of lan- guages whose similarity is intuitively realized by every student of Semitics should provide many specific features from which a reliable net of classifica- tory criteria could be worked out. According to a wide concensus, the principal method by which genetic classification of Semitic has to be guided is that of shared morphological innovations. This method, elaborated in a series of important contributions by R. Hetzron (e. g., 1972:13 and passim), was successfully applied by him and his successors to demonstrate the unity of West Semitic as opposed to East Semitic (Akkadian) and the unity of Central Semitic as opposed to other West Semitic languages. In spite of some difficulties connected with the for- mal shape of the corresponding classificatory features2 and the fact that con- flicting isoglosses have been adduced by some authors,3 the essence of Hetzron’s method proved to be valid and there is no doubt that it should be applied whenever possible — i. e., whenever important morphological inno- vations can be discovered for a given sub-branch of Semitic. Is this the case for Ethiopian Semitic? Regrettably, the answer seems to be overwhelmingly negative. As stated in Faber 1997:12, «although virtually all discussions of Semitic subgrouping assume a single Ethiopian Semitic branch which later split into North Ethiopic and South Ethiopic, there is virtually no linguistic evidence4 for such a Common Ethiopian stage». Indeed, even if Geez alone is confronted with Arabic, Sabaic or Mehri, reliable morpholo- gical innovations separating it from these languages are rather difficult to find, and the more so if such innovations are expected to be shared by, e. g., Tigre, Tigrinya and Amharic. Faber’s claim5 is perhaps exaggerated since 2 The origin of the second -a- in the Common WS New Perfect *katal-a³³ and that of -u in the Common CS New Imperfect *ya-ktul-u³³³ are still uncertain (HUEHNERGARD 2002:126 is well aware of this difficulty). 3 The principal one consists in the fact that Arabic shares several specific patterns of broken plural with Ethiopian and MSA (RATCLIFFE 1998:120). Cf. also ZABORSKI 1991, GOLDENBERG 1998:298ff. 4 Within Faber’s approach, «linguistic evidence» is clearly synonymous with «shared morphological innovations». This implicit equation — independent of its relevance for the subgrouping question — is terminologically inaccurate as it auto- matically excludes phonology, lexicon and syntax from linguistics. 5 No doubt, a fully justified reaction against many attempts to collect various allegedly specific Ethiopian features, all or most of which are actually absent from several languages of the group (notably, Geez), or turn out to be obvious retentions from PS, or can be qualified as typologically trivial developments possibly of areal nature. A typical example of such a list is LESLAU 1975 (as recognized already by Ullendorff in his comments on Leslau’s communication). HETZRON 1972:17–19 pro- vides a far more serious attempt but the final outcome is also somewhat disappoint- Downloaded from Brill.com10/02/2021 09:12:28PM via free access L. Kogan 369 some morphological features which can be treated as Proto-Ethiopian can probably be detected. I mean first of all the gerund base *k³atîlÐ-, shared by Geez, Tigrinya, Amharic and Argobba (on the controversial evaluation of the Gurage evidence v. Hetzron 1972:101ff. as opposed to Goldenberg 1998:466ff.); the agent noun *k³atâl-îy- (partly replacing the PS active parti- ciple *k³³âtil-) and the infinitive in -ot can also be mentioned. However, none of these features appears to be of crucial value for determining the linguistic unity of ES (note that at least in one language — modern Harari — all of them are lacking) and, even if they do occur together (as in Geez), they hard- ly correspond to one’s expectations about a group of languages intuitively perceived as closely related.6 Does the lack of specific morphological innovations mean that the lin- guistic unity of Ethiopian Semitic cannot be demonstrated? Should we under- stand ES as a geographical and historical-cultural concept only? In my opi- nion, one important factor should prevent one from drawing such a conclu- sion, namely the fundamental unity of the basic vocabulary. A systematic demonstration of this unity and its key relevance for the question of the com- mon origin of ES is the principal goal of the present article. Vocabulary is overtly or tacitly assumed to be persona non grata in most studies in the genetic classification of Semitic. According to Renfroe 1992:7, «that, under any circumstances, vocabulary is an unreliable measure by which to determine the relationship between any two given languages» is a «widely recognized fact». My comparativist’s intuition (no doubt shared by many col- leagues from various fields of historical linguistics) suggests that this — cer- tainly unreflected — attempt to deny the classificatory value of the lexicon cannot be correct. The vocabulary can and should be taken into consideration ing. Thus, verbal expressions based on the verb ‘to see’ are scarcely attested in Geez but so common in Cushitic that their wide use in Modern Ethiopian may well be regarded as an areal phenomenon affecting each language independently. As for some syntactic phenomena discussed by Hetzron, the concluding statement of the corres- ponding section of his book is quite telling : «...the Cushitic evidence came later and was independent in the different branches of Ethiopian». In sum, it is very hard to agree with APPLEYARD 1996:207–208 who believes that Hetzron’s study «lays to rest the phantom of a dual or even multiple origin of Ethiopian Semitic». Personally, I am convinced that Hetzron’s attempt to defend the common origin of ES is among the weakest points of his otherwise brilliant monograph. 6 Prof. R.-M. Voigt kindly reminded me in personal communication about one Proto-Ethiopian innovation overlooked by Faber and myself, namely the front vowel after the first radical in the imperfect of B (Gez. y#-sebb# r, Tna. y# -s#bb#r, Har. yi-sîbri). I can only object to his observation that even this (no doubt, very important) innovative feature is absent from Tigre where the imperfect of B is identical to that of A (l# -sabb#r). Incidentally, Tigre lacks gerund and preserves *k³âtil- as the main pat- tern of the active participle (though k³âtlây and kat³ âl(i) are also in evidence). Downloaded from Brill.com10/02/2021 09:12:28PM via free access 370 Scrinium I (2005). Varia Aethiopica in any subgouping procedure, especially if other criteria are scarce or vague (as in the case of ES). The main problem consists in elaborating a sound methodology for a proper assessment of its role. As far as the purpose of the present contribution is concerned, a simple and reliable method is at hand, in its essence almost symmetrically parallel to that used by Hetzron and his followers.