Sie sind auf Seite 1von 36

O print from: Babel und Bibel 6: Annual of Ancient Near Eastern, Old Testament, and Semitic Studies Copyright

t 2012 Russian State University for the Humanities. All rights reserved.

Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic*


Maria Bulakh
Russian State University for the Humanities, Moscow

1. Introduction One of the isoglosses separating neatly the North Ethio-Semitic (NES) and the South Ethio-Semitic (SES) languages is the shape of the general negative marker: !ay-/!i- in North Ethio-Semitic vs. *!al- in South EthioSemitic (Faber 1997:12). The former element is the basic negative morpheme in Geez, Tigre and Tigrinya, and its remnants can be discerned, in a few fossilized forms, also in SES. Similarly, the main SES negative marker *!al- is found marginally also in NES. Both markers have cognates outside Ethio-Semitic (ES) and are traceable to proto-Semitic. At rst sight, it may seem reasonable to assume that both of them were present in proto-Ethio-Semitic and, whereas the former ousted the latter in North Ethio-Semitic, the opposite development took place in South EthioSemitic. Such an interpretation would imply, incidentally, that *!ay-/*!i- as a general negative marker is a shared innovation of NES and perhaps the only signicant argument in favor of NES as a genealogical unity (Faber 1997:12; BulakhKogan 2010:280281). However, a closer inspection of the facts reveals that the reconstruction presented above is an oversimplication: the actual relationship between *!al- and !ay-/!i- can be obscured by the alternation -l-/-y-, well attested in SES. Besides, the distribution between the two negative markers in proto-ES should be more strictly dened. Are we indeed faced with mere free variants? Such a reconstruction is not impossible, of course, but is better avoided before other, more meaningful possibilities are fully exploited and rejected. It was probably these or similar considerations that led G. Hudson (2003) to question the traditional understanding of the synchronic and
* I am deeply grateful to Dr. L. Kogan who discussed the text with me in detail and helped improve both its content and style. Warm thanks go to Dr. S. Loesov who also carefully read the manuscript and made several insightful comments. Needless to say, I am responsible for all errors.

386

Articles: Semitic Studies

diachronic distribution of the negative particles in ES and to offer a new interpretation of the negative verbal forms in SES. According to the traditional view (v. references in Hudson 2003:209), reflexes of *!al- in SES are employed in negative forms of both perfect and imperfect. Within Hudsons new analysis, the negative *!al- is used in SES only in the perfect, whereas in the imperfect paradigm *!ay- rather than *!al- is employed. The negative marker *!ay- is thus reconstructed for the imperfect/jussive, and *!al- for the perfect (Hudson 2003:217). This analysis has important implications for the reconstruction of the negative markers of proto-ES. It shows SES to be more heterogeneous than NESand therefore, more likely to have preserved the original, proto-ES, negative paradigm. It seems, however, that some of Hudsons claims are in need of a detailed examination, whereas some of the facts to which he drew attention can be evaluated in a different way. A few arguments supporting Hudsons new analysis and a few other which contradict it are discussed below in this contribution, together with some additional evidence. Special attention will be paid to the negative forms of existential verbs and copulas whichas correctly pointed out by Hudsonmay have preserved archaic negative elements which are lost elsewhere.1 * * * Table 1 describes the negation in ES in the following contexts (each of them is morphologically distinct from the others in at least one ES language): negation of verbal forms (he did not do, he does not do), negation of non-verbal forms (not doing, not good etc.), negative form of the copula (he is not tall), negation of the locative verb (he is not there), negation of the predicative possessive construction (he does not have), negation of the existential construction (there is not, referred to as presence in Hudson 2003). For each position, data are given from a number of languages, which represent each of the major ES branches.2
For the importance of these forms for the reconstruction of proto-ES negative markers v. already Cohen 1931:3435. 2 The data are drawn from the following grammatical descriptions: Raz 1983; Elias 2005; Simeone-Senelle 2008; Dillmann 1907; Leslau 1941; Voigt 1977; Leslau 1995; Leslau 1997; Cerulli 1936; Wagner 1983; Leslau 1958; Meyer 2005; Meyer 2006; Gutt 1997; Leslau 1956; Leslau 1968; Leslau 1981; Leslau 2004; Hetzron 1977; ChamoraHetzron 2000. I am deeply grateful to Dr. A. Wetter who kindly shared with me the relevant paradigms for Argobba of onke/ollahaa variety of Argobba in several respects more conservative and less close to Amharic than the one described by Leslau (v. Wetter 2006).
1

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

387

In some languages, the patterns of negation in different dialects are not identical. This is why Tigre of Mensa (the dialect described by Sh. Raz, referred to as Tigre throughout the article) is treated separately from Tigre of Habab and Dahalik (whose statusa dialect of Tigre or a separate languageis debatable), and Argobba of Aliyu Amba from Argobba of ollaha. Some general remarks on various processes influencing the negative paradigms in ES are now in order. In Tigrinya, some Tigre dialects and most of SES, the predicative forms with negative prexes in the main clause are expanded with sufxed *-n (Tigrinya,3 Tigre of Habab) or *-m (SES), so that negation is marked by a circumx rather then a prex.4 The second element of these circumxes is of little relevance for the present discussion. Therefore, for the sake of simplicity the negative preformatives are referred to as markers of negation throughout this article, be they independent morphemes or rst elements of circumxes. In a number of languages, the paradigm of the imperfect of the main clause differs from that of the subordinate clause. Not infrequently, the subordinate forms preserve archaic negative markers, whereas in the paradigm of the main clause innovative means of expressing negation are used (as in Wolane, Selti, Gafat, Soddo). The negative markers used with nominal forms in Tigrinya, Amharic, Harari, Zway and Wolane are preceded (in Amharic, Zway and Wolane, optionally) by a relative pronoun. In Argobba of ollaha (in the perfect) and in Modern Harari (in the relative imperfect) the base of the jussive is employed in the negative
With dialectal variants -ni and -(3)y (Voigt 1977:236). In Modern Harari, the second part of the circumx (-(u)m) is sufxed to the verb in the perfect (al-sbr-m), but precedes the reflex of the negative auxiliary in the complex imperfect (y3sabr-um-l < *y3sabr-um-!i-hal, BREAK:3MS:IMPFNEG-NEG-AUX). Interestingly, in many languages the second part of the circumx can sometimes be attached not to the verb, but rather to another part of the sentence: Amh. l3u al-ma-m the child did not come vs. l3u-m al-ma the child did not come (in the second sentence the element -m functions as a kind of focus marker, cf. Leslau 1995:293 where -m in such contexts is analyzed as a conjunction of insistence). Similarly, in Modern Harari the element -m in the negative copula can be sufxed to the copula (al-ta-m), or to the preceding word (...-m al-ta), or, more rarely, to any other word in the negative sentence (Garad Wagner 1998:89). For the Hamasen dialect of Tigrinya, Voigt (1977:237) records a negative marker ny-, with the element n- occupying the position before the verb and merging with the prex !ay-.
4 3

388

Articles: Semitic Studies

forms. This is undoubtedly due to the influence of Oromo and/or Somali (Wagner 1997:598599). Likewise, in Zway some types of verbal conjugation employ the base of the jussive for the negative perfect. * * * It is important to distinguish between lexically restricted negative markersconned to one or several particular rootsand general negative markers. The latter term will be applied below to the negative morphemes which function as the principal means of expressing the negation in this or that ES language. A language may have more than one general negative marker, in which case they are either interchangeable or follow some distributional patterns. In a given ES language different general negative markers can be used with verbs and nouns or with different types of verbal forms. Lexically restricted negative markers in ES are typically associated with lexemes functioning as equational copulas, locative verbs, exponents of predicative possession, and existential words. Within a given language, morphosyntactic devices expressing equation, location, possession and existence may overlap: an equational copula can be used in locative constructions;5 a locative verb can shift to existential word; possessive constructions can be based on a locative verb or a locative prepositional phrase6 and can also be a source of existential constructions. In the majority of ES languages locative, possessive and existential constructions (henceforth referred to as LPE) are based on one and the same element, going back to the locative verb *hallawa, as in Amh. all he is (there)/there is; allw he has (lit. it is on/at him). Alternatively, possessive and existential constructions can be based on prepositional phrases with *ba- in, the reflex of *hallawa being restricted to locative constructions: Gez. bo he has/there is vs. hallo/hallawa he is (there). It turns out that LPE constructions based on prepositional phrases are more widespread in the negative forms than in the afrmative ones. Thus, the reflex of *hallawa is employed as the universal LPE afrmative predicate in Tigrinya, but its negative counterpart y-llo-n does not appear in the possessive constructions: instead, the form y-billu-n he does
5 Such a use is indeed found in ES (Amh. nkarra n I am strong and ahun bet n I am at home at present, cf. Leslau 1995:271, 273). However, this phenomenon was disregarded in the present study as irrelevant for the discussion. 6 For the locative constructions as a source of possessive constructions v. Heine 1997:5053.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

389

not have is used, going back to a possessive element based on the locative preposition *ba- (cf. fn. 21). It is crucial for the present investigation that the distribution of negative markers in LPE constructions normally depends on the lexical element to which the negative marker in question is bound rather than on the particular semantic function of the construction as a whole. For instance, the negative morpheme *!i- is regularly connected to the reflexes of *hallawa regardless of its concrete LPE function (v. 3.2 below). Conversely, the preposition *ba- in negative constructions is practically always combined with the negative reflexes of the marker *!al(a)-, sometimes with further markers imposed on it (v. 3.1 below).7 It stands to reason that lexically restricted negative markers may represent archaic morphemes ousted elsewhere by innovative general marker(s). Correct understanding of their evolution is thus of paramount importance for a reliable reconstruction of the negative markers of protoES (cf. fn. 1). The evidence presented above makes it clear that the apparent functional shifts affecting lexically restricted negative markers (e. g., from locative to existential) are in fact conditioned by the semantic evolution of the lexical elements associated with them. This means that the functional load of a reconstructed general negative marker cannot be deduced from the syntactical contexts in which its lexically restricted continuation is attested in this or that individual ES language. Furthermore, the distribution of various negative markers in a given ES language usually depends on the morphological shape of the negated word rather than on the semantics of the negative construction as a whole, and there are good reasons to believe that the situation in proto-ES was quite similar. It is therefore more important to establish for each negative marker the morphological contexts in which it could be used in proto-ES. For a lexically restricted marker, the morphological form of the corresponding lexical element, if traced back to proto-ES, can help us make some valuable conclusions about the respective negative marker. Such conclusions, however, can by no means be considered exhaustive. For instance, if a negative morpheme is restricted throughout ES to one parLPE constructions based on the same lexical element but using different negative markers are a rarity throughout ES. One characteristic example is Tna. yl-bo-n there is no vs. y-billu-n he does not have: both go back to the negation of the prepositional phrase with ba- in, but the negative prex, synchronically, is yl- (< *!ay-!al-) in the rst case and simply y- (< *!ay-) in the second one (v. fn. 21).
7

390

Articles: Semitic Studies

ticular verbal lexeme in the perfect, it is likely to go back to a proto-ES negative element which could negate verbal forms in the perfect. However, the reconstructed marker need not be conned to this morphological context and may have had a broader distribution for which no material evidence has reached us so far. It seems convenient to begin my presentation with the analysis of the general negative markers, to be followed by the discussion of the lexically restricted ones. The corresponding results will then be confronted to each other and a general reconstruction of the negative paradigm in proto-ES will be proposed. 2. General negative markers in ES 2.1. General negative markers in Tigre, Geez and Tigrinya As mentioned above, Tigre and Geez employ the negative marker !i- to negate all verbal forms. The same is true of the negative marker !ay- (...-n) in Tigrinya. Non-verbal forms (nouns, adjectives, innitives, etc.) do not require any special negative morpheme. In Tigre and Geez, they are negated with the general marker !i-. In a similar way, Tigrinya employs the general marker !ay- (...-n) with adjectives and innitives (Leslau 1941:74, 94), sometimes expanded with a relative pronoun (zy- < z3- + !ay-) (Leslau 1941:94, TED 2017). The general negative marker in Tigre of Habab is y3- (...-n), which appears in free variation with !i- except for the jussive, where only !i- is possible (Elias 2005:175). For Dahalik, Simeone-Senelle records (y)i- (...-ni) as a general negative marker throughout the paradigm. In the Tigrinya variety of May Tchew, y- (...-n) rather then !ay- (...-n) is used as the marker of negation (Voigt 2006:896). There is hardly any doubt that all the NES negative markers mentioned above are traceable to a single proto-ES morpheme, in its turn related to comparable morphemes elsewhere in Semitic (Brockelmann 1908:500 and, more recently, Wagner 1964:265, Faber 1991:414).8 Generalization of the negative marker !ay-/!i- (to the detriment of !al-) has been proposed as a rare example of a shared innovation of NES (Faber 1997:12).9
Reliable cognates are Akk. ay/ (AHw. 23, GAG 81i), Pho. !y (FriedrichRllig 1999:178), Hbr. ! (sparsely attested; HALOT 38, Ja 43), Soq. ! (LS 47). 9 Derivation of the Tigrinya negative marker !ay- from *!al-, ventured by Praetorius (1871:242) and supported by Hetzron (1972:21), has little to commend itself since -l- is not normally palatalized into -y- in Tigrinya.
8

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

391

The exact path of development which led to the emergence of the variants !ay-, !i-, y3-, (y)i- and y- is still in need of precision. To begin with, it seems likely that forms in y- go back to one of the !forms. More concretely, derivation from *!i- provides the easiest way of explanation. As will be mentioned below (section 3.2), in Tigre of Mensa the negative marker !i- has a special allomorph y3- before verbal forms with initial gutturals. One may suspect that this allomorph, initially restricted to this specic environment, was later generalized at the expense of the more original !i-. This explanation would also account for the vowel (instead of the expected 3) in Tigrinya of May Tchew: free alternation between a, and 3 before gutturals is characteristic of Tigrinya (Leslau 1941:12). As for *!i- and *!ay-, both are well compatible with the common Semitic variants mentioned in fn. 8 and may have coexisted in proto-ES, perhaps within a distributional pattern similar to the one observable in Epigraphic Geez. 2.1.1. Negative paradigm in Epigraphic Geez (EG) In the early inscriptions, both !i- (with its allomorph !3-, v. fn. 55) and *!ayare used. In this respect, Epigraphic Geez must have been close to proto-ES, whereas in Tigre, Classical Geez and Tigrinya paradigmatic leveling took place, !i- being generalized in Tigre and Classical Geez, and !ay-, in Tigrinya. Table 2 presents the attestations of the negative markers found in the vocalized inscriptions10 (in these examples, the transliteration of RIE is preserved, which means that consonantal gemination is not marked; phonological transcription is given in square brackets whenever necessary). This table reveals two facts. Firstly, the negative !ay- is attested only before the prex y3- (3 masc. sg. imperfect/jussive).11 Secondly, it is only in one position (3 masc. sg. imperfect) that both !i- and !ay- are found. Moreover, in the early inscriptions the variant !i- occurs only once in this
In the non-vocalized inscriptions, the negative marker appears as !-, followed by the 3 masc. sg. prex of the imperfect or the jussive. Such attestations are obviously not informative for the present discussion. For the same reason I leave out of consideration the negative marker d!, marginally attested in the epigraphic corpus. 11 One could even claim that the negative particle is not !ay- but rather !a-, attributing the glide y to the verbal prex: !a-y(3)um NEG-3MS:JUSS:stand. Such an analysis would, however, contradict the comparative evidence: the negative element *!ay- is convincingly reconstructed for proto-ES and could well be preserved in Epigraphic Geez, whereas there are no reasons whatsoever either for reconstructing a negative marker *!a- in proto-ES or for assuming the change *!ay- > *!a- in Epigraphic Geez.
10

392

Articles: Semitic Studies

morphological context (RIE 187:4; less important is the evidence from RIE 232:911, which belongs to a later period), against !ay- registered thrice. This makes the hypothesis of free variation between !i- and !ayundesirableone cannot fail to notice that !ay- is preferred before the prex y3- of 3 masc. sg. imperfect/jussive, and !i- elsewhere (nothing can be said, of course, about those verbal forms whose combinations with negative morphemes are not documented in EG, such as 2 pers. imperfect/jussive). In principle, an even stricter distribution between !i- and !ay- could be postulated, restricting the latter to only one position, viz. 3 masc. sg. of the jussive. To make feasible such a hypothesis, one would have to interpret y3tmaww! in RIE 188:5, RIE 189:4, and RIE 189:6 as jussive rather than imperfect. Such a usage in itself is unproblematic: at least once in the epigraphic corpus (RIE 192B:4) the jussive is actually employed in a similar construction (l-!gzbr / l--d! / ymt / w-!-ybl = *la-!3gzi<!a>ber la-za-da! y3mut wa-!ay-y3bli to God, may he not die or grow old (lit. who may not die or grow old)). In this passage, the unvocalized form ymt can only stay for the jussive (*y3mut). This interpretation would imply that in the phrase which is not to be conquered by the enemy the verb was variably used in the imperfect (RIE 187:4) or the jussive (elsewhere). Such a variation is, however, rather unlikely within such a clearly stereotype expression: a slight vocalic change in the negative prex is much easier to imagine. The negative !ay- seems thus to be primarily connected with the verbal prex y3- of 3 masc. sg. imperfect/jussive. It is tempting to see in EG a transitory stage from the original negative paradigm with a rigid distribution between !ay- (before y3-) and !i- (elsewhere) to that of Classical Geez where !i- is generalized in all positions. The most likely explanation for this distribution is that the allomorph !ay- is a result of dissimilation of -i- before -y3- (!i-y3-> !ay-y3-)12. It seems likely, therefore, that the emergence of the allomorph !aywas conditioned by the following sequence -y3-. This assumption allows us to reconstruct this marker for each position of the verbal paradigm where y3- is present, viz. 3 masc. sg. and 3 masc. and fem. pl. imperfect/jussive (the pertinent negative forms are not attested in the EG corpus). However, already in the early Geez inscriptions !i- began to penetrate even this position. In later periods, the allomorph !ay- was not in use any more.
12 On the sequence *-y and various means of avoiding it in Semitic v. Brockelmann 1908:250251.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

393

It is possible that such a distribution had existed already in proto-ES. On a later stage, the allomorph !i- was generalized in Classical Geez and Tigre, and the allomorph !ay-, in Tigrinya. 2.2. General negative markers in SES 2.2.1. Negative markers used with verbs Traditionally, the negative marker *!al- is thought to underlie all negative verbal forms throughout SES and hence reconstructed as the only general negative marker for proto-SES (Cohen 1931:35, Hetzron 1972: 21). Within an alternative approach developed by Hudson (2003), the negative marker *!al- in SES is present only in the perfect, whereas in the imperfect/jussive the underlying negative marker is *!ay-. To be sure, the reconstruction of the negative marker *!ay- for the imperfect/jussive in SES is not improbable. Indeed, in the SES negative paradigm of the imperfect/jussive the element -l- surfaces in 1 sg. only (Amh. al-sbr-3m I do not break), being allegedly assimilated to the following consonant elsewhere: at(t3)-sbr-3m (2 masc. sg.), at(t3)-sbri-m (2 fem. sg.), ay-sbr-3m (3 masc. sg.), at(t3)-sbr-3m (3 fem. sg.), an(n3)-sbr-3m (1 pl.), at(t3)-sbru-m (2 pl.), ay-sbru-m (3 pl.). More precisely, it is usually claimed that *-l- is palatalized into -y- before *y3- (with a subsequent simplication of -yy3- into -y-, cf. Brockelmann 1908:262) and assimilated to t- and n- in *t3- and *n3- with a subsequent (optional) degemination and syncope. In 1 sg., the development is assumed to be *!al-!3- > *al3- > al-, i. e. loss of the gutturals in agreement with the general trend of SES and subsequent syncope of -3-. As far as the 2/3 person sg./pl. and the 1 person pl. forms are concerned, the choice between *!al- and *!ay- in the underlying form can indeed be considered debatable: there is no immediate reason for the traditional *al-t- > at- to be preferred to Hudsons *ay-t- > at-. But the 1 sg. form *!al- is, at rst sight at least, by far less ambiguous, and it is this form that provides the most serious argument in favor of the traditional reconstruction and against Hudsons thesis. Being aware of this difculty, Hudson rejects the traditional understanding of -l- in the negated forms of 1 sg. as part of the negative morpheme *!al-. Instead, he considers it as a marker of 1 sg., identical to the element -l- in the 1 sg. prex of the jussive (Amh. l3sbr let me break).13
For the SES element -l- in the 1 sg. prex of the jussive and its parallels elsewhere in Semitic v. Huehnergard 1983:580.
13

394

Articles: Semitic Studies

Hudson does not deny that l in the jussive need not immediately bear on the diachronic interpretation of the negation in the imperfect. He nevertheless observes that in some SES languages (notably in Argobba) the element l is present also in 1 sg. and pl. of the afrmative imperfect: Arg. 3lsk3r I am drunk.14 For Hudson (2003:212), the element -l- in both the jussive and the imperfect is a reflex of the proto-Semitic asseverative particle la-, extensively discussed in Huehnergard 1983. An immediate objection to Hudsons hypothesis is that l-forms in the afrmative imperfectunlike l in the jussive and the negative imperfectare by no means common SES. In view of this circumstance, Hudsons hypothesis must force one to assume that the reflex of the asseverative *la-, originally present in the 1 sg. afrmative imperfect and jussive, was taken over to the negative imperfect but, for some reason, subsequently disappeared in the afrmative imperfect in most SES languages. Such a development is clearly very unlikely. Moreover, Hudsons claim that asseverative l ... was naturally favored in emphatic negative, 1st person is not compatible with the evidence of other Semitic languages, where the asseverative particle *la- was conned to main clause assertions, and did not co-occur with negative particles (Huehnergard 1983: 592). Thus, Hudsons reconstruction turns out to be highly articial in comparison with the more straightforward and economical traditional explanation which presumes that the negative particle *!al- was used in the imperfect as well as in the perfect. Hudsons other arguments against the traditional view (2003:212) are all concerned with the optional gemination of the consonant of the personal marker in 2 pers. and 1 pers. pl.: Amh. at(t3)-sbr-3m do not2 fem. sg. break, etc. As just mentioned above, within the traditional approach this gemination is thought to result from the assimilation of *-l- and the forms without gemination are explained as due to secondary simplication. Hudson, on the contrary, considers the geminated forms to be secondary, emerging in order to avoid the contact between the consonant of the prex and the rst radical which risked to result in full assimilation: atsb3r > *assb3r. Hudsons explanation is supported by the fact that secondary gemination of the prex consonant is attested elsewhere in the Amharic verbal paradigm: consider, for example, the optional gemination of -n- in the 1 pl. afrmative form (3n(n3)sb3r we break) or in the afrmative subordinate verbs (b3t(t3)sb3r if you break). However, this
14

After prexes also in Zway, cf. fn. 48.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

395

explanation undermines the reconstruction of *!ay- exactly in the same way as it does with *!al-. Indeed, if the forms with gemination are secondary, the last consonant of the negative marker must have disappeared without leaving a trace, which would be highly atypical for both -l- and -y-. [T]he reasonable loss of y before consonants posited by Hudson (2003:215) is difcult to accept in the absence of compelling parallel examples15. In short, the hypothetical shift *!ay-t3sb3r > *!atsb3r faces exactly the same difculties as *!al-t3sb3r > *!atsb3r. At the same time, Hudsons seven arguments in support of his own analysis (2003:213217) are by no means flawless. Three of these arguments rely on the general assumption that loss of -y- is more plausible that loss of -l-. For Hudson, fluctuating gemination in 2 pers. and 1 pers. pl. in Amharic (Argument 1) as well as absence of gemination in the corresponding forms in Gafat (Argument 5) and Selti (Argument 7) all suggest that the consonant of the negative morpheme was omitted rather than assimilated and, consequently, point to *!ayrather than *!al- in proto-SES. As pointed out above, this reasoning is questionable: simplication of gemination is at least as plausible as complete and traceless loss of -y-. Argument 2 focuses on the prexes ay- and - used with the negative copula (*dbl) in Amharic and Soddo respectively, as well as on the prex a- before the perfect forms of *kona in the Argobba variety of Aliyu Amba. As will be shown below (3.3), the Amharic and Soddo negative copulas are 3 masc. sg. verbal forms whose conjugation originally followed the paradigm of the imperfect rather than that of the perfect. In such a context, the use of *!ay- is only to be expected. As for the prex a- in Argobba, it can be reasonably traced back to *!al- (v. below, 3.4). Argument 3 derives from Hudsons important observation that the negative marker *!i- is not entirely absent from SES, but can be detected in the paradigm of the locative verb *hallawa. Within Hudsons approach, the hypothetic *!ay- in the imperfect represents still another (and by far more broadly present) SES remnant of the common ES negative marker *!ay-/*!i-. For Hudson, the presence of the negative marker *!i- in the perfect of *hallawa favors the presence of *!ay- in the imperfect of all verbal lexemes in SES. However, the usage of *!i- in combination with
Hypothetic monophtongization of *ay into e cannot be relevant as it would involve change of the vowel quality, which is not the case. Nor does the variation ay ~ a in aydllm adllm he is not provide a good analogy: the unique phonetic processes affecting this form are undoubtedly conditioned by its high frequency.
15

396

Articles: Semitic Studies

*hallawa is a clear example of a lexically restricted negative morpheme and may well have functioned as such already in proto-SES, where as the cumulative evidence of the attested SES languages plainly suggests all other verbs employed the general marker *!al-. In other words, the presence of the negative *!i- in the paradigm of one single verb is an important piece of evidence for the proto-ES reconstruction, but it does not tell us anything on the distribution of general negative markers in SES. In his Argument 4, Hudson claims that the prex an- (instead of the expected al-) in the 1 sg. negative imperfect/jussive in Harari (coinciding with the 1 pl. form) can only result from the combination of the negative a- (ay-) and the reflex of asseverative *l3- (> n3-). In his opinion, the shift l > n in Harari is restricted to the 1 sg. afrmative jussive (n3sbr < *l3sbr). When a similar shift is observed in 1 sg. negative imperfect, it means, for Hudson, that we are faced with one and the same morphological element. However, sporadic change from l to n is found in various contexts throughout ES (cf. Podolsky 1991:48 for Amharic), and its occurrence in two etymologically unrelated morphemes is not so improbable as Hudson believes. At the same time, merger of 1 sg. and 1 pl. in the negative imperfect paradigms in some ES languages is a matter of fact which can hardly be denied: note the negative subordinate paradigm in Selti, where both 1 sg. and 1 pl. display -l- rather than -n-: al-nakt I do not beat, al-naktina we do not beat (Gutt 1997:923). The same considerations apply to the 1 sg. negative prex in Soddo (ann3-). Argument 6 deals with the main-verb negative imperfect/jussive paradigms of Soddo and Gafat and the presumably related afrmative paradigm of the subordinate verb in Zway. Although these three paradigms display the element -l- in 1 sg. vs. gemination of t/n in 2 and 1 pl., Hudson convincingly shows that there is no negative morpheme *!al- in the background of such forms. This fact, however, is no argument for reconstructing *!ay- as the negative marker in proto-SES: in each of these languages, the common SES negative paradigm is also in evidence (in Soddo and Gafat restricted to the subordinate clauses). None of the arguments brought forward by Hudson is thus solid enough to postulate *!ay- (rather than *!al-) as the underlying form of the negative marker in the imperfect/jussive paradigm of SES. However, still another potential piece of evidence in favor of *!ay- in at least some of the imperfect forms might be detected in the Argobba variety of ollaha.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic sg. assm"3m att3sm"3m att3sm"3m ism"3m att3sm"3m pl. ann3sm"3m att3sm"um ism"um

397

1 2 masc. 2 fem. 3 masc. 3 fem.

The NEG:3MS/PL prex in the imperfect is i-, opposed to aC- elsewhere. Does this prex i- show that the underlying morpheme is *!ay-/!irather than *!al-? Such a proposal, however tempting, must nevertheless be rejected. As long as one accepts Hudsons reconstruction of *!ay- as the general negative marker in the imperfect/jussive in proto-SES, the cumulative evidence of all SES languages except Argobba of ollaha would point to *!ay- rather than *!i- in the prototype. Moreover, the Argobba picture can hardly be considered archaic: the hypothetic distribution of *!ay- and *!iin the Argobba imperfect paradigm does not match the distribution observed in the corresponding paradigm of Epigraphic Geez (2.1.1), but is rather exactly the opposite: Argobba shows i- in the only position where EG has *!ay-, and displays aC- (potentially traceable to *!ay-) elsewhere, (against *!i- in EG). The i-vowel in Argobba of ollaha must therefore be considered a secondary development from *!ay-.16 As for the underlying consonant of the negative morpheme aC- in the other forms of the imperfect, it could be either *-l- or *-y-: for both of these phonemes, assimilation to the adjacent consonant is easy to imagine
16 Such a development could be tentatively explained by the necessity to avoid the homonymy between the NEG:1S and NEG:3SM morphemes. Since assimilation of the consonantal element of the negative marker in Argobba of ollaha favored the evolution of *!ay- into aC-, the resulting paradigm would have lacked any morphological distinction between 1 sg. and 3 sg. masc. The shift from *!ay- to *!i- blocked the assimilation and made impossible this undesirable development. A similar reasoning could be applied to the negative paradigm of the imperfect/jussive in Epigraphic Geez (2.1.1), which, as just observed above, structurally resembles the negative paradigm of Argobba of ollaha. In Geez, the negative prex !i- triggers the dissimilation of the 1 sg. prex !3- into y3-, which becomes homonymous with the 3 sg. imperfect/jussive prex y3- (!i-y3sabb3r I do not break/he does not break). Since there are good reasons to believe that this phonological rule was operative in early Geez, the use of a special negative allomorph in 3 masc. sg. could be regarded as an attempt to avoid the ambiguousness of forms like !i-y3sabb3r. However, the very fact that homonymy between 1 sg. and 3 masc. sg. in these forms is tolerated in Classical Geez speaks against this hypothesis.

398

Articles: Semitic Studies

in this variety of Argobba, where the tendency towards assimilation of the consonantal element of the negative marker seems to be stronger than anywhere in ES. There is nothing to suggest that the negative prexes in the imperfect of proto-Argobba were to any serious extent different from those found in Argobba of Aliyu Amha or in Amharic, that is, *!al- for NEG:1S and *!ay- for NEG:3SM. As soon as this conclusion is accepted, one has to acknowledge that the evidence of Argobba of ollaha cannot be used as an argument in favor of *!ay-/*!i- as the general negative marker of the imperfect in proto-SES. And, conversely, diachronic explanations proposed for the emergence of the negative morphemes in the imperfect paradigm of Amharic turn out to be valid also for proto-Argobba: in both languages, the NEG:3MS element *!ay- can, as we have seen, result from either *!al-y3- or *!ay-y3-. Summing up, the evidence of the SES languages does not provide any persuasive argument in favor of reconstructing *!ay- as the imperfect/jussive negative marker in proto-SES. Rather, the element *!al-, surfacing in 1 sg. throughout SES, speaks against such a reconstruction. However, in view of the evidence of Epigraphic Geez the picture turns out to be more complicated. In section 2.1.1 above I have tried to reconstruct for Epigraphic Geez (more tentatively, also for proto-ES) a complementary distribution between the allomorphs *!i-/*!ay-, the latter appearing before the personal prex *y3-, and the former elsewhere in the verbal paradigm. Theoretically, such a distribution can also be postulated for proto-SES, assuming that the allomorph *!i- was replaced by *!al-, whereas *!ay- before *y3remained unchanged. In other words, the element *!ay- (NEG:3MS/PL) in the imperfect/jussive in SES would go back to proto-SES *!ay-y3-, whereas for the rest of the verbal paradigm the negative marker *!alwould be reconstructed. Such a hypothesis, while not wholly improbable, can hardly be proved and is deemed to remain a purely theoretical construct.17
The negative paradigm of Argobba of ollaha may be thought to support this reconstruction (the distribution between the allomorphs aC- and i- is exactly the same as one between *!i- and *!ay- reconstructed for Epigraphic Geez), but in fact it does not exactly for the same reasons which have prevented us from using it to support Hudsons reconstruction: one is forced to assume then that the allomorph *!i- was replaced with *!al- in proto-SES, whereas the change from *!ayto *!i- in 3 masc. sg./pl. took place in Argobba of ollaha only. The latter step,
17

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

399

A few other general negative markers (such as !3l-, !3lla- in East Gurage or t- in Gafat and Soddo) are innovative morphemes restricted to just one or several ES languages and have no relevance for the reconstruction of the negative markers of proto-ES. 2.2.2. Negative markers used with non-verbal forms A peculiarity of at least some of the SES languages (unfortunately, for many idioms of the subgroup no exact information on this part of grammar is available) is the employment of a special negative marker for negating non-verbal forms (nouns, adjectives, verbal nouns).18 Transparent manifestations of this phenomenon are extant in Amharic, Zway and Wolane.19 From the synchronic point of view, it is only in these languages that non-verbal forms are negated with a special marker *!ala- (which can be optionally expanded with the relative pronoun *ya-). In Harari and Selti, the morpheme *!al- is used with both verbal and non-verbal forms, in the latter case always preceded by the relative pronoun *z3- in Harari. However, the data of Amharic, Zway and Wolane seem to suggest that the pre-nominal negative element *!al- has only secondarily become homonymous with the pre-verbal *!al- and may also go back to the same prototype *!ala-. The syncope of a short unstressed a in an open syllable is not improbable, although I have to admit that I am not aware of a fully reliable comparable example of such a phenomenon in ES. Still, omission of a short -a- in the course of a syllable structure rearrangement is considerably easier to imagine than (apparently, nonmotivated) insertion of -a- independently in several languages. Some evihowever, is equally compatible with the element *!ay- (NEG:3MS/PL) going back to *!al-y3-. 18 For Tigrinya, Leslau (1941:94) reports a special negative marker zy- before innitives. The underlying negative element is still *!ay- (*z3- being a relative pronoun), by no means different from the general negative marker of Tigrinya. 19 Nominal lexemes comprising the negative markers !al-/!ay- in Zway, such as !aly ignorant < !al-l-u he did not know, !aysoru granary entrance/aperture < !ay-sor-u he does not steal (Meyer 2005:247) are not considered here since their negative markers are taken over from the verbal negative paradigm. For deverbal negative adjectives in Selti a phonological distribution between !ay- and !al- has been reported by Gutt (1997:910), the latter form appearing only before vowels (ayno sterile, barren < e give birth, al-sbo unexpected < saba notice). One wonders whether this distribution is a further development from a system documented in Zway or whether !ay-/!al- in Selti continue the negative marker *!al(a)- of non-verbal forms (for the possibility of such a development v. fn. 36).

400

Articles: Semitic Studies

dence in support of this reconstruction can be found in the lexically restricted negative *!al(a)- (3.1), which can be identied with the negative morpheme under scrutiny. 3. Lexically restricted negative markers In the discussion of the negative markers occurring with copulas, locative verbs and possessive constructions, attention should be paid to the specic verbs or predicative elements with which the lexically restricted negative markers are associated. At least in some cases, the combination of a given negative prex with this or that root is ubiquitous in ES and can be safely reconstructed for the proto-language. Negative forms of the pertinent predicative elements are examined below in this section, in an attempt at establishing for each of them its original negative marker. 3.1. Possessive expressions involving *ba + possessive sufx are negated with the particle *!al(a)- in NES (Geez, Tigre of Mensa, Tigre of Habab, Dahalik, Tigrinya) as well as in SES (Gafat). An important question is, however, which of the variants is original: *!ala- (Tigre of Mensa, Tigre of Habab, Dahalik, Gafat) or *!al- (Geez and Tigrinya)? The primacy of the former element appears more attractive: the shift *!alabo > *!albo in Geez and Tigrinya can be explained as a result of vocalic syncope (v. 2.2.2), whereas gemination of -l- in Dahalik (y-alle-b-ni) and Gafat (allb-am) can be regarded as strengthening of the rst syllable in order to avoid the syncope. Furthermore, the reconstructed negative *!ala- is compatible with the negative marker *!ala- combined with nominal forms in some of the SES languages (2.2.2).20 Conversely, transition from *!al- to *!ala- is hardly plausible as there is no immediate reason for the hypothetic insertion of -a- between !al- and -bo. In Dahalik, Tigrinya and Gafat the old negative possessive construction has been expanded with new negative morphemes, a process obviously to be accounted for by desemantization of *!al(a)-. In Dahalik, the negative construction is augmented with the general negative marker y-(...-ni): y-alle-b-ni he does not have. As for Tigrinya, an extensive discussion of the whole spectrum of the expanded forms with a special attention to their historical background can be found in Voigt 1977:237 250. For the present purpose, it is sufcient to say that the most common
20 For this comparison v., e. g., Leslau 1956:149. Note that in the case of the pre-nominal *!ala- the second vowel is absent (presumably syncopated) in Harari and Selti.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

401

form of the negation of existential construction is *!albo augmented with the general negative marker *!ay--n: *!ay-!albo-n > yl-bo-n.21 In Gafat, the element -am (also a part of the general negative circumx) is imposed upon the old possessive construction. Another peculiarity of this language is that *!alab- has been reanalyzed as a verbal stem and is conjugated through the sufxes of the perfect. 3.2. Negative forms based on *hallawa in the expressions of location, possession, or existence are negated with !i- in Tigre and Geez. It seems reasonable to assume (with Hudson 2003:213214; v. already Praetorius 1879:259) that the corresponding forms in the rest of ES (Tna. y-llo-n, Amh. y-ll-m and, with y- > e//, Har. l, Zwy. !l(l)o, Cha. en, etc.) go back to the same proto-form. The shift from !i- to y- in Tigrinya and SES can be plausibly explained by a morphophonemic rule !VH- > yVH-, synchronically operative in the negated forms of verbs with initial gutturals in Tigre (*!i-azze > y3-azze I do not want/he does not want).22 One can also surmise that the negative markers y3- and y- recorded for some NES idioms (v. 2.1) result from generalization of the allomorphs of *!ibefore gutturals. 3.3. Negative forms of the copula based on the root *dbl are found in Amharic, Gafat and Soddo.23 In Soddo (Leslau 1968:13), all forms of the negative copula consist of the preformative , the stem of the imperfect
21 The negative form of the possessive construction (ybillu-n) may also go back to the combination *!al(a)- + ba- + pronominal sufx. It can be explained as a result of metathesis of l and b with a subsequent reinterpretation of l as an applicative sufx (hence, possibly, its gemination). Alternatively, it can be seen as a further development (simplication) of the rarely attested yl-b3llu-n, clearly derived from *!al-b- + pronominal sufx and expanded with a general negative marker and the applicative sufx -l- (for a detailed analysis v. Voigt 1977:242 243). In Voigts view, however, ybillu-n rather goes back to a regular negative form of the otherwise unattested possessive construction (a combination of the preposition !ab-, the applicative -l- and a pronominal sufx), v. Voigt 1977:247. 22 Saleh Mahmud Idris, communication made in the framework of the International Workshop History and Language of the Tigre-speaking Peoples (Eritrea and Sudan), Naples, 78 February 2008. A similar diachronic rule probably accounts for such Amharic forms as yaz to take < *!aaza, although one has to admit that in some non-verbal roots containing two adjacent gutturals (such as *!aad- one or *!ut- sister) no change !- > y- is registered anywhere in ES. 23 Leslau (1956:77) mentions the form adb3l for Chaha, which is, however, missing from EDG I and EDG as well as from Hetzron 1977. Note also the obviously related expression aldbl except, with the exception of, found in Soddo and Wolane (EDG 39).

402

Articles: Semitic Studies

and an object pronoun (with the exception of the 1 person, where the sufx of the perfect conjugation is used instead of the object pronoun).24
3 masc. 3 fem. 2.masc. 2 fem. 1 Sg. -db3ll (-IMPF:OBJ.3MS) -db3lla (-IMPF:OBJ.3FS) -db3kk (-IMPF:OBJ.2MS) -db3 (-IMPF:OBJ.2FS)25 -dbukk (-PERF:SUBJ.1S) Pl. -db3llm (-IMPF:OBJ.3MP) -db3llma (-IMPF:OBJ.3FP) -db3kk3m (-IMPF:OBJ.2MP) -db3kk3ma (-IMPF:OBJ.2FP) -db3lln (-IMPF:OBJ.1P)

The paradigm of the Gafat cognate is very similar except for the addition of the innovative morpheme of the negative imperfect t--am (Leslau 1956:77).26
3 masc. 3 fem. 2 masc. 2 fem. 1 Sg. t--db3ll-am (t--IMPF:OBJ.3MS-am) t--db3llat-am (t--IMPF:OBJ.3FS-am) t--db3kk-am (t--IMPF:OBJ.2MS-am) t--db3-am (t--IMPF:OBJ.2MS-am)27 t--db3kkw-am (t--PERF:SUBJ.1S-am) Pl. t--db3llm-an (t--IMPF:OBJ.3P-am) t--db3kkmw-am (t--IMPF:OBJ.2P-am) t--db3lln-am (t--IMPF:OBJ.2P-am)

The verbal root *dbl is well known in ES with the meaning to add, repeat, join; parallels with the meaning to collect, unite are found in Arabic and Soqotri (v., e. g., EDG 195196; for its comparison with the negative copula v. Leslau 1956:77, Goldenberg 1968:74, Hudson 2003:213 214 as well as EDG 1415 for an overview of alternative proposals). The semantic development is reconstructed as follows: X is not added to Y > X is not counted as Y > X is not Y.28 The Amharic negative forms aydll-m, ay-doll-m is generally considered to go back to the same verbal
Note that this form has no Main Verb Markers, and that the object sufxes of the 2 person are attached to the stem in a way slightly different from that in which they are normally attached to the imperfect stems (namely, without a vowel inserted between the stem and the pronoun: y3gdl--h he kills yousg., y3gdl-3h3m he kills youpl.). 25 Normally, the 2 fem. sg. object pronoun is -, but the form - appears in the impersonal conjugation (Leslau 1968:2325). 26 As in Soddo, no vowel is inserted between the stem and the 2 pers. object pronoun, although such a vowel usually appears when the object pronoun is attached to a regular verb in the imperfect. The forms of the object sufxes of 2 fem. sg. and 2 pl. deviate from the sufxes employed with regular verbs (- and -hum < *-kum respectively). 27 In Gafat, both - and - are present as the exponents of the 2 fem. sg. object pronoun (Leslau 1956:5960). 28 Cf. the usage of the Wolane verb dl- in the meaning to be counted as or even to become in the expression tsb ydl 3l being circumcised, lit. a child which is added to the human beings (Meyer 2006:140).
24

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

403

root *dbl (with the shift *-ab- > *-aw- > -o-). Indeed, the Old Amharic evidence (Getachew Haile 1970:75) favors such a comparison, revealing both the original -o- and the object sufxes rather than the perfect subject sufxes attached to the base. Relying on the actual forms appearing in the old manuscripts, as well as on the comparative data from Gafat and Soddo, one can propose the following paradigm of the negative copula in Old Amharic.
3 masc. 3 fem. 2 masc. 2 fem. 1 Sg. !ay-dol(l)w-3m (*!al-IMPF:OBJ.3MS-m) !ay-dol(l)at-3m (*!al-IMPF:OBJ.3FS-m) !ay-dol(l)h-3m (*!al-IMPF:OBJ.2MS-m) !ay-dol(l)-3m (*!al-IMPF:OBJ.2FS-m) !ay-dol(l)hu-m (*!al-PERF:SUBJ.1S-m) Pl. !ay-dol(l)aw-3m (*!al-IMPF:OBJ.3P-m) !ay-dol(l)ahu-m (*!al-IMPF:OBJ.2P-m) !ay-dol(l)n-3m (*!al-IMPF:OBJ.1P-m)

This reconstructed paradigm differs from the one proposed in Getachew Haile 1970:75 in one position, viz. 1 sg.: the form *!aydoll3m I am not reconstructed by Getachew Haile is neither found in the manuscripts nor conrmed by data from Gafat and Soddo. It is reasonable to assume that the Old Amharic paradigm of the negative copula which immediately preceded the modern one was similar to that of Gafat and Soddo. Of course, this does not exclude the possibility that the paradigms of Old Amharic, Soddo and Gafat, in their turn, go back to a paradigm which had object sufxes in all positions (similar to the one offered by Getachew Haile). The substitution of the object pronominal sufxes with the perfect subject sufxes which one can thus observe in the history of the negative copula in Amharic is undoubtedly related to the similar replacement in the afrmative copula, postulated by Hetzron (1972:8081) for the majority of SES languages.29 Interestingly, this replacement did not occur simultaneously in the afrmative and negative copulas. As we have seen, in the negative copula based on the root *dbl this development took place in Amharic only. As for the the afrmative copula, it has preserved the original sufxes in Amharic, Argobba and Gafat, whereas in the rest of SES they were replaced with the perfect subject sufxes.30
In other words, the Amharic negative copula provides an additional argument in support of Hetzrons claim that the object sufxes in the copula are more archaic. 30 The use of subject (perfect) and object pronominal sufxes within one and the same paradigm of a negative copula is also registered for Argobba of Aliyu Amba (Leslau 1997:3536, v. below, 3.4), where the object sufxes are found in 3
29

404

Articles: Semitic Studies

It becomes clear that the subject sufxes of the perfect paradigm in modern Amharic represent a relatively recent innovation: originally, as suggested by the evidence of Gafat and Soddo, the morphological form of this copula was that of 3 masc. sg. of the imperfect, which agreed with the logical subject by means of the object pronoun. This reconstruction undermines Hudsons reasoning according to which the Amharic form ay-dll-m represents a rare (and, presumably, archaic) case of the negative element ay- combined with the perfect and provides a direct link between SES *!ay- (not traceable to *!al-) and the main negative marker !ayin Tigrinya (Hudson 2003:214). The data presented above suggest that the prex ay- in the Amharic negative copula is diachronically the regular Amharic negative prex of 3 masc. sg. imperfect: ay-dll-m/ay-doll-m is thus not different from ay-sbr-3m. In the corresponding forms of Gafat and Soddo the prex *!ay- does not surface, however: the shape of the base under scrutiny is db3l-31 and it can hardly be a coincidence that both languages display -- before the rst radical, which is difcult to explain synchronically. What is the origin of this vowel? For Hudson (2003:214), the initial element - in Soddo is a continuation of *a- < *ay-. Since the alleged shift *a- > - is difcult to explain, one may wonder whether the development *ay > *e > could provide a plausible alternative. Such a shift, however, would also be unique since the combination ay- is usually preserved in both Soddo (ay-rs he does not begin) and Gafat (ayfr3k he cannotsubordinate). An irregular simplication of *ay is not that improbable for a grammatical word of high frequency. Indeed, the Amharic form adllm is to be regarded as an independent example of a similar simplication from a more regular aydllm. We may conclude that the negative copula based on the root *dbl originally consisted of the imperfect base preceded by a vocalic element (probably going back to a combination of a personal prex of the imperfect and of a negative marker) and followed by an object sufx which agreed with the logical subject. This may be true for all forms of the negative copula, although in 1 sg. the object sufx is replaced by the subject sufx of the perfect in all attested paradigms.
sg. and pl. and the subject sufxes of the perfect elsewhere. In Argobba of ollaha, only subject sufxes are used (A. Wetter, p. c.). 31 The only difference between Gafat and Soddo is that the Gafat form also displays the new negative prex t-, obviously imposed on the older negative copula as soon as its original negative marker became desemantized.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

405

3.4. Negative copulas based on the verb *kona usually attach general negative markers rather than any lexically restricted markers. The only exception is Argobba of Aliyu Amba, where the negative copula has the form a-huney-u, apparently with a negative prex a- instead of the expected general marker al- or the preformative ay- (NEG:3MS). Hudson (2003:214) seems to interpret this prex as a continuation of the negative marker *!ay- (rather than *!al-), but the data from Argobba of ollaha permit one to reconsider this question from a different angle. Indeed, the paradigms of the negative copula of Argobba of Aliyu Amba and Argobba of ollaha are structurally similar: the base of the jussive is followed by an object sufx in 3 masc. sg. and pl. and by a subject sufx of the perfect in the rest of the paradigm. In Argobba of ollaha, the negative marker used with this combination is *!al- (-l- sometimes surfacing, but mostly assimilated to the rst radical of the verb), in agreement with the general pattern of negation of the perfect conjugation in this idiom. It stands to reason that the element a- in the negative copula of Argobba of Aliyu Amba must go back to the same negative morpheme, *l being rst assimilated to the initial radical and then dropped as a result of simplication of geminated consonants (v. fn. 40). 3.5. It is generally believed that the indeclinable negative copula !akko in Geez derives from the verb kona to be, become combined with the negative marker *!al- (CDG 14 with further references). This etymology seems to be supported by the Tigre negative copula !i-kon which, as the Geez one, is indeclinable, but preserves the nal -n and is thus clearly related to the verb kona. At the same time, it is hard to refrain from comparing Geez !akko with Zway !unku and Selti inko, very similar to it in both form and function.32 The vowels of the rst syllable in East Gurage point to original *3,33 which makes it difcult to derive !un-/in- from *!al-. Instead, these elements are to be identied with *!3n-, a marginally attested but doubtless archaic negative marker found in such fossilized forms as Gez. !3n-d"i, Tna. !3n-d3"i, Amh. 3n-a I do not know, Gez. !3n-bi, !3n-b3ya no, Tna. !3m-bi bl refuse (Leslau 1969b:140).34 If this etymology is correct, one has to choose between *!3n-ko and *!al-ko as two alternative proto-ES reconstructions.
32 The form !unku in Zway is probably to be explained by a mutual assimilation of the rst and second vowels (*!3n-ko > !unku). 33 For the variation between 3, i and u in Zway v. Meyer 2005:46. In Gutts transcription of Selti, i is the regular reflex of 3. 34 Leslau further compares the element 3n-, combined with 2 person forms of the perfect and expressing prohibition in Muher, Mesqan, and Western Gurage,

406

Articles: Semitic Studies

The replacement of *!3n- by *!al- is indeed not improbable since *!al(a)- in proto-ES apparently functioned as the general negative marker in the same positions where *!3n- probably occurred as a lexically restricted one: before nouns (*!3n-d") and before the preposition *ba(*!3n-ba+POSS.1S). In such a context, identication of *-ko with kona may seem a major problem: why a verbal form should have used negative markers typically associated with non-verbal forms? Still, one can surmise that the use of non-verbal negative markers in this case was conditioned by desemantization of -ko, which already in proto-ES probably began to be perceived as a non-verbal form. The emergence of negative copulas based on kona in Tigre and elsewhere in ES must then be attributed to independent development. 3.6. It is not to be excluded that the negative copula in Harari (alta-m) and the negative locative verb in Argobba of ollaha (yata-m) are related. The element -t- in the Harari form is certainly the same as in the positive copula (with the basis t-, v. GaradWagner 1998:45). Given the fact that there is no plausible diachronic explanation for Argobba yata-m, one may hypothesize that it is based on the same element -t-, followed by sufxes of the perfect and probably preceded by the general negative marker *!al-, which, as in some other positions (fn. 38 and 40), could be assimilated to -t- with the subsequent loss of gemination. The negative locative verb in Argobba of ollaha is thus structurally identical to the negative copula of Harari (for the shift from copula to locative verb, v. section 1 with fn. 5). One must admit that the initial y- in Argobba of ollaha remains difcult to explain. One might tentatively analyze it as a relative pronoun, whose appearance with a negative copula or a locative verb would, however, be quite unusual. 4. Reconstruction of the proto-ES negative paradigm 4.1. Table 3 (q. v.) gives a summary of the evidence presented and analyzed above. A comparison between general negative markers and lexically restricted negative markers allows one to draw two conclusions about proto-ES negative markers combined with the perfect and with nonverbal forms.
but this is rather unlikely. Instead, the Gunnn-Gurage 3n- can be compared to !3l(a)- as the negative marker of the main imperfect in Wolane and Selti.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

407

(1) The element *!i- can be reconstructed as the proto-ES general negative marker of the perfect. This function is preserved in Tigre of Mensa and in Geez, whereas in the rest of ES its traces can be detected in the reflexes of the negative forms of the locative verb *hallawa (which, as is well known, has the inflectional paradigm of the perfect when it refers to the present time). (2) Proto-ES probably used different methods for negating verbal forms and other words. The negative marker of non-verbal forms can be reconstructed as *!ala-, which preserves this function in some of the SES languages, but is restricted to negative constructions based on the preposition *ba- in Gafat, Tigre, Tigrinya and Geez. 4.2. My hypotheses concerning the proto-ES negative markers of the imperfect/jussive are considerably more speculative since the data allow several interpretations. The general negative marker of the imperfect/jussive is !ay- in Tigrinya and !i- in Classical Geez and Tigre, whereas in Epigraphic Geez a complementary distribution between !ay- and !i- is observed. As far as proto-SES is concerned, the general negative marker of the imperfect/jussive is to be reconstructed *!al- or, alternatively, a complimentary distribution between *!al- and *!ay- can be posited, structurally resembling the Epigraphic Geez picture. The most straightforward reconstruction based on these data is to assume that the distribution between !ay- and !i- in Epigraphic Geez faithfully reflects the proto-ES heritage. In such a case, various generalizations in the verbal paradigm must have occurred in Tigre, Classical Geez and Tigrinya, whereas in SES the prex *!al- (presumably going back to the proto-ES negative marker of non-verbal forms *!ala-) must have replaced the allomorph *!i- (or the morpheme *!i-/*!ay- in general) throughout the verbal paradigm. This reconstruction may seem to face some difculties. It is indeed hard to imagine how the proto-ES negative marker *!ala-, thought to be restricted to the rather marginal function of negating non-verbal forms in proto-ES, could have been generalized as the main negative morpheme in SES. Still, if one takes into consideration some phonological processes which likely affected the morphemes *!ay- and *!i- on their way from proto-ES to SES, the emergence of *!al- in SES can nd a satisfactory explanation. The development of the negative paradigm from proto-ES to protoSES is thought to be influenced by two major factors.

408

Articles: Semitic Studies

The rst one is the morphophonemic rule !VH- > yVH- which, as shown above (3.2), can be postulated for Tigrinya and SES. The negative marker was thus likely realized as *y3- before the forms of the perfect of verbs with initial gutturals as well as before the 1 sg. prex !3- of the imperfect/jussive. In the latter case at least, such a realization must have brought about the clearly undesirable development *!i-!3- > *y3!3- > *y3-, with the resulting homonymy of 1 sg. of the negative imperfect/jussive with 3 masc. sg. of the afrmative imperfect/jussive. It may well be that generalization of !ay- at the expense of *!i- in Tigrinya was triggered by the necessity to avoid this homonymy. One has to reconstruct a similar generalization for proto-SES in order to account for the further development: the replacement of *!ay- by *!al- is likely to be attributed to a phonological process (for which v. immediately below), whereas a direct change from *!i- to *!al- can hardly be explained in phonological terms. The second factor potentially accounting for the shift from *!ay- to *!al- in SES is the alternation between -l- and -y- a phenomenon well attested in SES, but conspicuously absent from Geez, Tigre and Tigrinya. Indeed, palatalization of *-l- into -y- before front vowels and y (as in Amh. by sayimv. fem. sg. < *bl-i) is a regular morphophonemic process in SES. The opposite development, viz. hardening of the original intervocalic -y- into -l-, can thus be regarded as hypercorrect interpretation of each intervocalic -y- as resulting from palatalization. This kind of hypercorrection is admittedly rare and sporadic, but several reliable examples can be detected in some SES languages: Amh. abbl to break ones promise, promise falsely, Har. abla to refuse as opposed to Gez. !abaya to refuse35 or Amh. lla to be red vs. Gez. ea (*y) to be red (CDG 456, Cohen 1939:28, Praetorius 1879:48, 75; Podolsky 1991:36). A particularly interesting case is Gaf. al brother/al3t sister (Leslau 1956:175), likely traceable to PS *!a(w)- (* > *y > l). One can assume that the same hypercorrect development took place whenever the negative *!ay- was followed by a vowel. Although such environments in proto-SES were probably not so numerous as they are in most modern SES languages because of the loss of gutturals, there are good reasons to suppose that the shift was at work at least before the etymological *-!-, likely dropped earlier than other gutturals. The path of the development can be reconstructed as follows: *!ay-!V- > *!ay-V- > *!al-V-. The shift from *!ay- to
35

Contra Cerulli (1936:406) and Leslau (EDH 17), who reject this comparison.

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic

409

*!al- was especially prominent in the 1 sg. imperfect/jussive (which otherwise would have become homonymous with 3 sg. masc. imperfect/jussive) and in the dependent verbal forms (where the exponent of the negative marker was least discernible). Indeed, the shift from -y- to -l- is quite conspicuous in some dependent negative forms of the locative verb, cf. the Ancient Harari relative form of l (zi-ll or za-ll which is not < *z3-yll), Amharic y-lell (< * y-yll), Argobba of Aliyu Amba ylella (< * y-yll). The same process may be responsible for the emergence of the allomorphs y-/-l- of the 1 sg. prex in Zway (v. fn. 48). It is thus likely that *!al- in the verbal negative paradigm of SES emerged as a phonetically conditioned allomorph of *!ay-. How can one explain its expansion to preconsonantal environments? The existence of the phonetically similar old negative morpheme *!ala-, appearing before all non-verbal forms, including participles and innitives, must have contributed in no small way to this expansion. Generalization of *!al- can thus be regarded as a merger of two negative markers of different origin *!ala-, an archaic element used with non-verbal forms, and *!al-, an innovative allomorph of *!ay-.36 The present reconstruction does not presuppose a common NES generalization of *!i-/*!ay- as the principal negative marker throughout the verbal paradigm. Rather, generalizations of *!i- in Tigre and Classical Geez and that of *!ay- in Tigrinya are seen as independent developments from a common proto-ES stage, at which *!i- and *!ay- functioned as complimentary distributed allomorphs within the verbal paradigm. Moreover, generalization of *!ay- is reconstructed for proto-SES as well, as a stage preceding generalization of *!al-.37 In a similar vein, loss of the negative marker *!ala- conned to nonverbal elements appears as a trivial development which took place independently in Tigre, Tigrinya and Geez (most probably, in a few SES languages as well) and, as such, has no classicatory value. Conversely, generalization of *!al- as the general negative marker of the verb in SES turns out to be a reliable common innovation.
36 It is possible that the morphophonemic distribution of !ay- and !al- in negative deverbal adjectives in Selti (v. fn. 19) is due to a similar merger of the preverbal *!al-/!ay- with the negative marker *!al(a)- of the non-verbal forms. Obviously, the process in Selti should be dated at a more recent stage. 37 In view of the absence of *!ay- in Tigrinya of May Tchew it seems wise to refrain from treating *!ay- as a common innovation of Tigrinya and proto-SES.

410

Articles: Semitic Studies

This reconstruction is in agreement with other pieces of comparative evidence connected with the NES/SES dichotomy: while common SES innovations are by no means a rarity, there are hardly any convincing examples of an innovative morphological feature shared by Geez, Tigre and Tigrinya (BulakhKogan 2010).

Table 1. Negation in Ethio-Semitic languages

The sign = shows that this context is not treated as a separate morphological category in the language in question; the sign ? shows that no exact data have been found in the grammatical descriptions available to the author. Whenever gemination of the adjacent consonant is involved, no attempt at reconstructing the consonant of the prex is undertaken, so that the negative morpheme is referred to as *!aC-. The forms are normally adduced as they are quoted in the grammars, with necessary adaptations of the transcription. The negative morphemes (or elements which result from the fusion of negative morphemes with personal prexes) are in bold.
Dahalik !i-sabara !ay-sbr-n he did not he did not break break al-sbbr-m he did not break !i-y3sabb3r !ay-y3sbb3r-3n ay-sbr-3m Gez. Tna. Amh.

Tgr. (Mensa)

Tgr. (Habab)

Perfect

!i-sabra he did not break

y3-sabra-n(ni)/ !i-sabra he did not break

Arg. of Aliyu Arg. of ollaha Amba al-sddba-w he a-s3m"a-m (< did not insult *!aC-s3m"a-m)38 he did not hear ay-sdb-u i-sm"-3m

Imperfect 3 masc. sg.

!i-sabb3r

y3-...-n(ni)/!i-39

Imperfect 1 sg.

!i-sabb3r

y3-...-n(ni)/!i-

(y)i-...-ni (yi-me!eki-ni he did not come to you) (y)i-...-ni (yi-!ammer-ni he does not know) ? !i-y3sabb3r !ay-sbb3r-3n al-sbr-3m

al-skr-u I do not get drunk

as-sm"-3m (< *!aC-sm"-3m)40

The negative form of the perfect in Argobba of ollaha employs the base of the jussive, although the sufxes are still those of the perfect (Wetter 2006:902). It is unclear why gemination disappears in this form (an analogy with the corresponding form of the imperfect?). In other forms of the negative perfect paradigm, the rst radical is always geminated: ass3m"ewm (1 sg.), ass3m"ex3m (2 masc. sg.). 39 For the distribution between y3- ... -n(ni) and !i- v. section 2.1. 40 The prex al- surfaces in some verbs (al-m-3m I will not come, Wetter 2006:906). Forms with the negative prex a- and non-geminated rst consonant adduced in Wetter 2006:902 have been later re-analyzed by Dr. Wetter as having an underlying *al(A. Wetter, p. c.). Indeed, degemination of originally doubled consonants seems to be relatively common in this variety of Argobba.

38

Tgr. (Mensa) ? !i-/!akko + zy- + non-verb. inf./adjective form al-/yal- + non-verbal form ?

Tgr. (Habab)

Dahalik

Gez.

Tna.

Amh.

Arg. of ollaha

Negative non- !i- + nonverbal forms verbal form

Arg. of Aliyu Amba ?

Negative form of the copula y-elle-ni (?) !i-hallawa y-llo-n41/ yl-bo-n42 y-billu-n y-llw-3m

!i-kon (3 pers.), !i- + copula (elsewhere) !i-halla i-t-ni ay-dll-m/ ay-doll-m/ a-dll-m/ a-doll y-ll-m

!i-kon/!i-kon-i/ !i-kon-ini (3 pers.); y(3)- + cop. (1 & 2 pers.) !i-halla/ya-hallanni ? !al-bo

!akko (in!ay-kon-n decl.)/!i-/!ikona

negative forms of the reflexes of *kona: a-huney-u (y)ella-w (< *y-lla-w) yelley-u (< *y-lly-u)

neg. forms of the reflexes of *kona: ak-kuna-m (< *!aC-kuna-m) yat-a-m

!ala-bu

bu !i-kon/ !ala-bu

Negation of the locative verb Negation of poss. constr. (3 masc. sg.) Negation of existential construction y-alle-b-ni !al-bo yl-bo-n/ y-llo-n y-ll-m

!ala-bu

(y)ella-w (< *y-lla-w)

41

42

V. further variants in Voigt 1977:239. V. further variants in Voigt 1977:240.

Perfect

Modern Har. al-sbr-m he did not break ila-nakt ilaw-nakt ay-nakt

Sel. al-nakata he did not beat

Gaf. al-fttr-am he did not die45

Imperfect 3 masc. sg. Imperfect 1 sg. !al-n3l-u48 = !3ll3s-sbb3r (!3ll3C-sbb3r) -y-bl he does not eat

y3sabr-um-l 46

Zwy. Wol. !al-n3l-44 he !al-sbrm he did not pull did not break out !ay-n3l-u !3l3-sbb3r

Imperfect subordinate 3 masc. sg.

Ancient Har. al-borda he did not reach43 ay-sbri he does not break al-ur I do not go47 = ti-frk-am he can not til-frk-am/ t3l-frk-am ay-fr3k

isbr-um-

ay-sibr49

43

Wagner 1983:111, Text B, line 153; cf. Cerulli 1936:359. The negative perfect of this verb employs the base of the jussive (Meyer 2005:138, 192). There are, however, other types of verbal conjugation where the base of the afrmative perfect is the same as that of the negative perfect. 45 Leslau (1956:98) posits for the verbs with a nal - a negative circumx al-...-m (with a change of nal - to -a) and a negative circumx al-...-am for the rest of the verbs. It seems more economical to postulate a single morpheme al-...-am, with the nal - elided before -a- (for this rule v. ibid. 22, 11b). The origin of the element -a- in -am is unclear. 46 In Modern Harari, the negative forms of the imperfect in the main clauses go back to combinations of an afrmative imperfect form of the main verb with a negative form of the auxiliary verb (Wagner 1997:596). The negative forms of the simple imperfect, quoted in Leslau 1958:22, are, according to Wagner 1997:596597, not used in Modern Harari and either formed ad hoc by placing the negative prexes before the simple imperfect or the informants gave the forms which were known to them from liturgical texts. 47 Wagner 1983:127, Text B, line 222. 48 According to Meyer (2005:96), the 1 sg. imperfect prex in Zway has two allomorphs: y- in word-initial position and l- after other prexes (cf. Zwy. t-l-n3l-u while I take). Consequently, he does not regard -l- as part of negative prex !al-. This distribution, however, may be due to reanalysis of *!al- (NEG:1S) as !a-l- (NEG-1S), with l- perceived as an allomorph of 1 sg. prex in noninitial position (cf. section 4.2 for the shift -y- > -l-). The same explanation can be valid for the element -l- in 1 sg. after the negative marker t- in Gafat. In both cases the present authors analysis differs from that of Hudson (2003:216), for whom these two cases are clear examples of the original function of l(3)- as the marker of 1 sg. in the afrmative imperfect paradigm. 49 In the negative forms of the imperfect of the subordinate clauses in Modern Harari, the negative prexes are attached to the jussive rather than to the imperfect stem (Wagner 1997).

44

Zwy. = al-nakt al- + innitive50 ? t-db3ll-am51 inko (indecl.)/negative forms of reflexes of *kona: al-na la (< *y-la) lay (< *y-ly)

Ancient Har. Modern Har. Imperf. subordinate = an-sibr 1 sg. Negative non-verbal al- + innitive zal- + non-verbal form forms Negative form of the ? al-ta-m, m-al-ta copula

Wol. -l-bl

Sel.

Gaf. al-fr3k

l (< *y-l)

l (< *y-l)

all-b-am (?)52 all-b-3w-am54

Negation of the locative verb Negation of possessive construction !l(l)o (< *y-llo) !l (< *y-l) la (< *y-la)

l (< *y-l)

Negation of existen- l (< *y-l) tial construction

la-m/yela-m (< *y-la-m), l-ba-m/yel-ba-m (< *y-l-ba-m)53 l (< *y-l)

!al-/yl- + !al-/yl- + innitive non-verb. form !unku (indecl.) neg. forms of reflexes of *kona: !al-n !l(l)o !l (< *y-l) (< *y-llo) !ley !ley (< *y-ly) (< *y-ly)

all-b-am

50

V. Gutt 1997:962. For the prexes !al- and !ay- combined with deverbal adjectives and nouns in Zway and Selti v. fn. 19. From the morphological point of view, this is a verbal form of the imperfect followed by object sufxes and preceded by tinstead of the expected ti-/t3- (NEG:3MS, v. Leslau 1956:77). 52 Sufxes of the perfect are attached to the stem allb- (< *!ala-b-). 53 Negation of the locative verb with a pronominal sufx, with or without the locative sufx -b-. 54 The same base as in the negation of the locative verb, with a pronominal sufx inserted between the stem and the perfect ending of 3 masc. sg. The origin of the element -3-, appearing in forms of the 3 person, is unclear.

51

Perfect

Sod. al-sfr he did not measure

Enm. -spr-da (< *an-spr + da < !al-sbbr + da) he did not break

Perfect subordinate e-sb3r (< *!ay-sb3r) ann3-sb3r = = ? negative forms of the reflexes of *kona ? =

Muh. Msq. an-sbr an-sbr (< *!al-sbbr) he (< *!al-sbbr) he did not break did not break = =

Chah. an-tks (< *!al-tkks) he did not kindle =

Imperfect 3 masc. sg.

Imperfect 1 sg. = = ? neg. forms of the reflexes of *kona: an-xn (< *!alkn) ynn (< *y-ll) ? ? ? ynnn (< *y-lln) ynn (< *y-ll) ? ?

t-irs he does not begin t-rs

e-sb3r (< *!aysb3r) ann3-sb3r

ay-rs

-spr (< *an-spr < *!al-sbbr) e-r3b (< *!ay-rk3b) ay-s3r-ka he does not nd a-kft (< *!al-kft) -s3r-ka (< *an-s3r + ka < *!alI do not open sb3r + ka) = ay-s3r

ann3-rs

Impf. subord. 3 masc. sg. Impf. subordinate 1 sg. Negative non-verbal forms Negative form of the copula

-s3r (< *an-s3r < *!al-sb3r) ?

db3ll (db3l with object pronoun)

neg. forms of the re- negative forms of the reflexes of flexes of *kona: *kona: an-xr (< *!al-kn) -xZrZ-da (< *an-xZrZ + da < *!alkn + da) en (< *y-ll) en-da (< *y-ll + da) enn (< *y-lln) en-da (< *y-ll + da)

Negation of the locayell (< *y-ll) tive verb Negation of poss. con- yellnn str. (3 masc.sg.) (< *y-llnn)

Negation of existential yell (< *y-ll) construction

Table 2. Negative markers !i- and !ay- in Epigraphic Geez

3 sg. imperfect

*!ayRIE 188:56: za!ay3tmaw! / la[a]r which is not to be conquered by the enemy RIE 189:4: za!ay3tmaw! / laar id. RIE 189:6: za!ay3tmaw / laar id.

1 sg. imperfect

!iRIE 187:4: za!3y3tmaw / laar (< *za-!i-y3tmaww(!) laar) which is not to be conquered by the enemy55 RIE 232:911: zabala"a / 3gya / wasatiya / dam3ya / !iy33"3m / lamot (he) who ate my flesh and drunk my blood will not taste the death RIE 189:89: sobe / tamakaa / wa!iy3fal3s56 / !3mtakazi / y3be / !3[z][ba] / nob when the Noba people boasted and said: I will not withdraw from Takazi RIE 189:48: !3nza / !i!3"em3 / !3zba while I do not oppress the peoples ? RIE 189:2122: za!i[y]!am3r / w3lo whose number I do not know (the reading !i is not quite certain) RIE 189:6: 3dmeya / !ay3um57 / ar let the enemy not stand before me

3 masc. sg. jussive

1 pl. perfect

RIE 187:11: !imoan[hu] we did not put (him) in chains

3 masc. sg. perfect 3 masc. pl. perfect

RIE 189:13: sobe / l3!ik3yani / !isam"ani when, (even) after I have sent, he did not listen to me RIE 189:1615: wa!3mz3 / gwayu / wa!iomu and after that they fled and did not stop DAE 12:1415: ...!itarakuni... they did not leave me58

55 !3- is most likely to be explained as a variant of !i- before -y-. This type of variation (-3y-/-iy- and -3w-/-uw-) is well known from both Epigraphic Geez (v. E. Littmann in DAE, p. 80 and Dillmann 1907:80, 9899, 348) and early Geez manuscripts (v. Bausi 2005:158). 56 The form is homonymous with 3 masc. sg. However, the verb y3be is likely to introduce the direct speech, as do its cognates elsewhere in ES. 57 Or, rather, !ayum. 58 In RIE (193:15), most of this inscription is treated as illegible (this particular word being transcribed as .-t-.-kuni).

Table 3. Negative markers in Ethio-Semitic languages: a summary


Lexically restricted negative markers *!i-: throughout ES before the verb *hallawa (*!i- > y- in Tigrinya and SES)

General negative markers !i-: general negative marker (perfect, imperfect, nominal forms) in Tigre and Geez; in 3 masc. sg. of the imperfect also in Argobba of ollaha

y3-: general negative marker in Tigre of Habab (also y- in the Tigrinya variety of May Tchew); allomorph of !i- before verbal forms with initial guttural in contemporary Tigre of Mensa !ay-(...-n): general negative marker in Tigrinya (all verbal forms); allomorph of !i- before the prex y3- in Epigraphic Geez; in SES, most probably not present at all !al-(...-m): general negative marker for verbs in SES; -lsurfaces in the perfect as well as in 1 sg. of the imperfect/jussive; in Harari and Selti also registered before nonverbal forms *!ala-: general negative marker with nominal forms in Amharic, Zway and Wolane (the same function probably traceable to proto-SES)

*!ala-: negative marker before the preposition *ba- in possessive and existential construction (such as Tigre !alabu) and traceable to PES. Cf. also Geez !akko, perhaps to be analyzed as *!al- (< *!ala- ?) + -ko (< *kona to be) *!3n- in a few fossilized negative expressions (such as Geez !3n-d"i and !3n-b-) and possibly discernible in the East Gurage negative particle *!3nko

418

Articles: Semitic Studies

Language abbreviations
Amh. Amharic; Arg. Argobba; Chah. Chaha; EG Epigraphic Geez; Enm. Ennemor; ES Ethio-Semitic; Gaf. Gafat; Gez. Geez; Har. Harari; Msq. Mesqan; Muh. Muher; NES North Ethio-Semitic; PES Proto-Ethio-Semitic; Sel. Selti; SES South Ethio-Semitic; Sod. Soddo; Tgr. Tigre; Tna. Tigrinya; Wol. Wolane; Zwy. Zway.

References
Bausi 2005 Brockelmann 1908 BulakhKogan 2010 Bausi, A. Ancient Features of Ancient Ethiopic. Aethiopica 8:149169. Brockelmann, C. Grundriss der vergleichenden Grammatik der semitischen Sprachen. I. Laut- und Formenlehre. Berlin. Bulakh, M.; Kogan, L. The Genealogical Position of Tigre and the Problem of North Ethio-Semitic Unity. ZDMG 160:273302. Cerulli, E. Studi Etiopici. I. La lingua e la storia di Harar. Roma. Chamora, B.; Hetzron, R. Inor. Mnchen. Cohen, M. tudes dthiopien mridional. Paris. Cohen, M. Nouvelles tudes dthiopien mridional. Paris. Dillmann, A. Ethiopic Grammar. Ed. by C. Bezold, tr. with additions by J. A. Crichton. London (repr. 1974). Elias, D. Tigre of Habab. Short Grammar and Texts from the Rigbat People. Harvard University. PhD. thesis. Faber, A. The Diachronic Relationship between Negative and Interrogative Markers in Semitic. Kaye, A. S. (ed.). Semitic Studies in Honor of Wolf Leslau on the Occasion of His Eighty-fth Birthday. Wiesbaden. Pp. 411429. Faber, A. Genetic Subgrouping of the Semitic Languages. Hetzron, R. (ed.). The Semitic Languages. London. Pp. 315. Friedrich, J.; Rllig, W. Phnizisch-Punische Grammatik. 3. Auflage. Roma. Garad, A.; Wagner, E. Harari-Studien. Wiesbaden. Getachew Haile. Archaic Amharic Forms. Proceedings of the Third International Conference of Ethiopian Studies. Addis Ababa. Vol. 2. Pp. 6180. Goldenberg, G. K3stan3a. Studies in a Northern Gurage Language of Christians. OrSu 17:61102. Gutt, E.-A. Studies in the Phonology of Silti. Journal of Ethiopian Studies 16:3773. Gutt, E.-A. Concise Grammar of Silte. Gutt, E. H. M.; Hussein Mohammed. Silte-Amharic-English Dictionary (with Concise Grammar by Ernst-August Gutt). Addis Ababa. Pp. 895 957.

Cerulli 1936 ChamoraHetzron 2000 Cohen 1931 Cohen 1939 Dillmann 1907 Elias 2005 Faber 1991

Faber 1997 FriedrichRllig 1999 GaradWagner 1998 Getachew Haile 1970

Goldenberg 1968 Gutt 1983 Gutt 1997

M. Bulakh, Negative markers *!ay-, *!i- and *!al- in Ethio-Semitic Heine 1997 Hetzron 1970 Hetzron 1972 Hetzron 1977 Hudson 2003

419

Huehnergard 1983 Leslau 1941 Leslau 1956 Leslau 1958 Leslau 1968 Leslau 1969a

Leslau 1969b

Leslau 1981 Leslau 1995 Leslau 1997 Leslau 2004 Meyer 2005 Meyer 2006 Mller 1905 Podolsky 1991 Praetorius 1871 Praetorius 1879 Raz 1983

Heine, B. Possession: Cognitive Sources, Forces, and Grammaticalization. Cambridge. Hetzron, R. Review of Leslau 1968. JAOS 90:561581. Hetzron, R. Ethiopian Semitic. Studies in Classication. Manchester. Hetzron, R. The Gunnn-Gurage Languages. Napoli. Hudson, G. Ethiopian Semitic Negative Nonpast. Bender, M. L.; Takcs, G.; Appleyard, D. L. (eds.). Selected Comparative-Historical Afrasian Linguistic Studies, in Memory of Igor M. Diakonoff. Mnchen. Pp. 209218. Huehnergard, J. Asseverative *la and Hypothetical *lu/ law in Semitic. JAOS 103:569593. Leslau, W. Documents tigrigna (thiopien septentrional). Paris. Leslau, W. tude descriptive et comparative du Gafat (thiopien mridional). Paris. Leslau, W. The Verb in Harari (South Ethiopic). Berkeley Los Angeles. Leslau, W. Ethiopians Speak. Studies in Cultural Background. III. Soddo. BerkeleyLos Angeles. Leslau, W. Is There a Proto-Gurage? Proceedings of the International Conference on Semitic Studies Held in Jerusalem, 1923 July 1965. Jerusalem. Pp. 152171 (repr. in: Id. Gurage Studies, Collected Articles. Wiesbaden 1992. Pp. 226245). Leslau, W. The Negative Particle !in in Arabic and (!)3n in Ethiopic. Annali dellIstituto Universitario Orientale di Napoli 19:137145. Leslau, W. Ethiopians Speak. Studies in Cultural Background. Part IV. Muher. Wiesbaden. Leslau, W. Reference Grammar of Amharic. Wiesbaden. Leslau, W. Ethiopic Documents: Argobba. Grammar and Dictionary. Wiesbaden. Leslau, W. The Verb in Msqan as Compared with Other Gurage Dialects. Wiesbaden. Meyer, R. Das Zay. Deskriptive Grammatik einer Ostguragesprache (thiosemitisch). Kln. Meyer, R. Wolane. Descriptive Grammar of an East Gurage Language (Ethiosemitic). Kln. Mller, D. H. Die Mehri- und Soqori-Sprache. II. SoqoriTexte. Wien. Podolsky, B. Historical Phonetics of Amharic. Tel-Aviv. Praetorius, F. Grammatik der Tigriasprache in Abessinien. Halle. Praetorius, F. Die Amharische Sprache. Halle (repr. HildesheimNew York, 1970). Raz, Sh. Tigre Grammar and Texts. Malibu.

420 Simeone-Senelle 2008

Articles: Semitic Studies Simeone-Senelle, M.-C. The Specicity of the Dahalik Language within the Afro-Semitic Languages. Paper presented at the International Workshop History and Language of the Tigre-speaking Peoples (Eritrea and Sudan). Naples, 78 February 2008 [published as Simeone-Senelle, M.-C. The Specicity of the Dahalik Language within the Afro-Semitic Languages. Lusini, G. (ed.). History and Languages of the Tigre Speaking Peoples. Napoli, 2010. Pp. 127145]. Voigt, R. M. Das tigrinische Verbalsystem. Berlin. Voigt, R. M. Sdtigrinische Dialekte: das einfache und zusammengesetzte Prsens im Dialekt von May-w. Uhlig, S. (ed.). Proceedings of the XVth International Conference of Ethiopian Studies, Hamburg July 2025, 2003. Wiesbaden. Pp. 893898. Wagner, E. Der bergang von Fragewrtern zu Negationen in der semitischen Sprache. MIO 10:261274. Wagner, E. Harari-Texte in arabischer Schrift mit bersetzung und Kommentar. Wiesbaden. Wagner, E. The Negative Imperfect in Ancient and Modern Harari. Fukui Katsuyoshi; Eisei Kurimoto; Masayoshi Shigeta (eds.). Ethiopia in Broader Perspective. Papers of the XIIIth International Conference of Ethiopian Studies. Vol. I. Kyoto. Pp. 596600. Wetter, A. The Argobba of TollahaA Comparative Overview. Uhlig, S. (ed.). Proceedings of the XVth International Conference of Ethiopian Studies, Hamburg July 2025, 2003. Wiesbaden. Pp. 899907.

Voigt 1977 Voigt 2006

Wagner 1964 Wagner 1983 Wagner 1997

Wetter 2006

Das könnte Ihnen auch gefallen