Paper 2
Contents |
- 1. Introduction : Hypotheses A and B
- 2. Evidence for and against Hypothesis A
- 2.1 Literary Arabic
- 2.2 Biblical Hebrew
- 2.3 Ugaritic
- 2.4 Old Aramaic
- 2.5 Akkadian
- 2.6 Ge’ez
- 2.7 Epigraphic South Arabian
- 2.8 Modern South Arabian
- 3. Evidence for Hypothesis B
- 4. Hypothesis A Refuted?
- 5. GP Subjunctive Forms
- 6. Aspect in Egyptian Verb Forms
- 6.1 Introduction
- 6.2 Outline of Egyptian Verb Morphology
- 6.3 The Functions of the śdm.f and śdm.n.f Forms
- 7. Semitic Suffixing Forms
- Bibliography
- Bibliographical Abbreviations
- Footnotes
1. Introduction : Hypotheses A and B
1.1 Among the more interesting problems in Semitic diachronic linguistics is the question of how, from a presumed common original, the differing verbal systems of the various languages evolved. Prior to the recovery of Akkadian it was generally assumed that the system of Classical Arabic, in its apparent ‘purity’, most closely resembled the original, so that the systems in other languages – for example Hebrew and Aramaic - were taken to originate in the ‘Arabic = Semitic’ system. However the verbal system of Akkadian, in its antiquity and with its apparently unique form-function patterns, substantially weakened the ‘Arabic = Semitic’ hypothesis and poses problems which have not been satisfactorily resolved. Nevertheless, despite its inability to account for the Akkadian data the ‘Arabic = Semitic’ hypothesis continues to have its adherents, not least because the alternatives have their own shortcomings.1
1.2 The original number of terms in the Semitic aspect system, and their values, has long been the subject of investigation.2 But while the differing conclusions drawn by investigators to some extent reflect the apparently irreconcilable evidence provided by the languages, another major difficulty has been the failure to recognise the limitations of ‘induction’ as a method of reasoning in diachronic linguistics, a dificulty compounded by a certain lack of precision in the names and definitions applied to the terms proposed for the system.3 In an attempt to address the problems posed by the inductive method the present study utilises a ‘hypothetical-deductive’ approach, along the lines advocated by Popper, whereby a hypothesis is proposed which appears to fit the data, and evidence is then sought in the various languages to either support or weaken the hypothesis.4
1.3 Therefore as an initial hypothesis the set of aspect morphemes in Common Semitic is taken to have comprised at least three members, expressing elements which will be termed <singulative>, <nonsingulative> and <stative>, and defined as follows;
- <Singulative> aspect is taken to ‘encode’ an event which the writer or speaker views as having occurred once, or whose interest in an event that has not yet taken place is confined to a single potential occurrence. Whether the event has occurred, or will occur, more than once is taken not to be relevant;
- <Non-singulative> aspect is taken to encode all events which the user cannot or does not wish to encode as
. Such events may be perceived as occurring more than once (iterative) or more generally persisting (durative); - <Stative> aspect is taken to encode a situation that is viewed as unchanging, contrasting with an event - whether viewed singulatively or non-singulatively.
1.4 Events and states can be represented graphically as shown in Figure 1, from which it will be seen that there are two sub-types of event. In the first sub-type (Figure 1A), which could for example represent she cut, the final state (not cutting), stripped of all connotations and implications, is the same as the initial state, as far as the grammatical subject is concerned. In the other sub-type (1B), where the event is grammatically passive, the initial and final states are different. There is of course a potential logical relationship between the two sub-types which becomes more evident if clarify is substituted for cut, to give the parallelism:
he clarified x vs x became clear
Sections of the ‘event lines’ not included within the boundaries defining <singulative> would in principle be encoded by <stative> although, with the exception of the final state clear in figure 1B, these are not states which would normally be expressed in natural language.5
Figure 1 Schematic Representation of Singulative Aspect
1.5 For Figure 1A it should be emphasised that encoding <singulative> is taken to imply an absence of consequence or result, such that, for instance, this event could not be rendered by English she has cut. Contrast with this the hypothesis that the principal aspectual distinction was between ‘complete’ and ‘incomplete’ action, where the former cannot logically exclude she has cut.6 With regard to <non-singulative>, the simplest definition would be that it comprises an unmarked series of <singulative> events for, as already noted, in many contexts <non-singulative> may have ‘iterative’ or ‘durative’ connotation. On the other hand it would be quite possible for a language to have <iterative> and/or <durative> as members of its system of aspect, and it will later be argued that the evidence suggests that Common Semitic and Egyptian, and the language from which they are taken to descend, may indeed have possessed an element akin to <iterative>, which would of course have implications for the definition of <non-singulative>.
1.6 Comparative evidence suggests that <stative> aspect in Common Semitic and Egyptian was expressed by a form with suffixed subject pronouns, exemplified by the Akkadian permansive and Egyptian old perfective ; in this study such forms will be termed GS (G-form suffixing). However, the discussion that follows will largely be confined to the elements <singulative> and <non-singulative>, which are postulated (in the simplest case) to have been expressed in Common Semitic by prefixing G-forms (GP). Two hypotheses will be explored, the first of which (Hypothesis A) argues that Common Semitic exhibited the following two basic types of GP form:
- An ‘apocopate’ form (denoted GPA), displaying no overt aspect morpheme and having the general structure (for a 3fs form) *tinpur, *tinpir or *tinpar This form is conjectured to have expressed <singulative> aspect;
- An ‘extended’ form (GPE), much as the GPA form but with aspect morpheme suffixed either to the stem or to suffixed morphemes of gender/number. Conjectured to have expressed <non-singulative> aspect.
Hypothesis A is explored in Section 2. The second hypothesis (B), explored in Section 3, proposes that Common Semitic made use of a GPA form as defined above, but with <non-singulative> aspect expressed by a form with geminate second root morpheme, perhaps originating in a pre-Semitic reduplicated stem. This form is taken to have had the general structure (3fs) *tinappur, again with stem vowel variants in i and a, and will be denoted GPG.
2. Evidence for and against Hypothesis A
2.1 Literary Arabic
2.1.1 For strong roots, and leaving aside passive forms,7 apocopate GP forms in Arabic (majzūm) are constructed on templates (3ms) yaqbur, yaqbar or yaqbir. Such forms are clearly GPA in that, aside from gender/number morphemes suffixed to the 2fs, 3p, 2p and dual forms, no morpheme occurs after the final radical. GPA forms are used in the following contexts:8
- Non-declarative constructions, e.g. jussive and cohortative (Vol. II, §17(a));
- Negative declarative and imperative (Vol. II, §12, §17(b), §18);
- Protasis and apodosis of conditional sentences expressing ‘possible’ conditions (Vol. II, §17(c)).9
None of these expresses a singulative event which has definitely occurred ; this function is fulfilled in Arabic by the GS form, which could be seen as at least a partial refuation of Hypothesis A (on the GS form in Semitic refer to Section 7). But all three attested functions can be accommodated by the definition of <singulative> proposed in §1.3 above, albeit with varying degrees of confidence ; for instance the third application is somewhat problematic in that such general statements would at first sight appear to fit better with the definition of <non-singulative>. However, although conditional constructions in the individual languages display many idiosyncracies, the particular type expressing ‘possible’ conditions appears to be Common Semitic, for it also occurs at least in Hebrew and Akkadian (see below at §2.2 and §2.5). This particular use of the GPA form is thus certainly ancient, and may have evolved through viewing such events as hypothetical, rather than specific, single events10
2.1.2 For the moment ignoring ‘subjunctive’ forms (Section 5 below), there are two Arabic G-forms which are ‘extended’ in the sense defined in §1.3, namely the muḍāri‘ and ‘energic’ forms, the second of which consists of two sub-types. Active voice muḍāri‘ forms other than 2fs, 3p and 2p are built (for strong roots) on templates yaqburu, yaqbiru or yaqbaru, and thus differ from their majzūm equivalents in adding final -u to the stem. The 2fs, 3mp and 2mp forms also add final –na, such that the 3fp and 2fp muḍāri‘ endings are identical to those of the equivalent majzūm forms. The muḍāri‘ is used in the following contexts:11
- To signal a non-hypothetical situation where no particular event or state is envisaged;
- When an event will occur in the future;
- In conjunction with selected verbs in their māḍī (GS) form to express:
- (a) An event considered to be ongoing in past time;
- (b) An event iterated in past time;
- (c) The purpose of a singulative event in past time.
2.1.3 Energic forms other than 2fs, 3mp and 2mp add either -an or –anna to the majzūm form.12 The 2fs, 2mp and 3mp forms shorten final ī and ū of the majzūm and reduce –anna to –nna ; the 3fp and 2fp forms are fairly regular, except that the sequence –na-anna becomes nānni.13 Energic forms are typically used under circumstances 1 and 2 above and commonly serve to emphasise the idea expressed by the equivalent muḍāri‘ form. But these forms also occur in certain kinds of conditional construction, commands, prohibitions, etc.,14 although with the exception of conditional contexts the foregoing functions can all be readily reconciled with an original aspect term <non-singulative>.
2.1.4 ‘Subjunctive’ 3p and 2p forms are identical to those of the equivalent majzūm forms so that, by comparison with the muḍāri‘, the subjunctive can be said to display an attenuated GPE paradigm. This fact, together with the evidence from Ge’ez (§2.6 below) and the absence of clear evidence for a morphologically analogous form in the other Semitic languages - with the very questionable exceptions of Ugaritic and the el-Amarna letters,15 suggests that the Arabic subjunctive may have been an internal innovation originating in the majzūm, and thus perhaps of no great relevance for the history of GP forms in Common Semitic.16
2.2 Biblical Hebrew
2.2.1 Leaving aside the Masoretic pointing, Hebrew GPA forms can most readily be detected in forms on II-weak and III-weak roots, which in certain environments tend to be written without the second or third radical respectively. Such forms occur in the following contexts:17
- Definitely occurring, usually singulative, events in past-time narrative (waw-consecutive construction);
- Certain non-declarative constructions, e.g. jussive and cohortative;
- Negative imperative constructions;
- Conditional constructions relating to possible conditions (GKC, §109 h/I, §159).
Function 1 has no parallel in Arabic but obviously offers quite good support for hypothesis ‘GPA ⇔ <singulative>’. Functions 2 to 4 are shared with the equivalent Arabic forms and support or weaken the hypothesis to the same degree as the Arabic data.18 But as in Arabic, a GS form expressing <singulative> aspect is common in Hebrew, and could again constitute a partial refutation of Hypothesis A.
2.2.2 If it is correct to view the relevant Hebrew II-weak and III-weak GP forms as structurally apocopate, it follows that equivalent forms where the weak radical is written may originate in GPE forms even though, at first sight, there appears to be no evidence for a morpheme after the third radical other than those of number and gender. There are however two phenomena in Hebrew that provide more solid evidence for hypothesis ‘GPE ⇔ <non-singulative>’. The first of these is a sporadic but not too uncommon final n attached to 2fs, 3mp and 2mp forms, although the first of these is much less common than the others. This n seems morphologically equivalent to the -na suffixed to the equivalent Arabic muḍāri‘ forms, is more common in poetry and is not infrequently paralleled by forms without n.19
2.2.3 The second phenomenon also involves phoneme n, but this time prefixed to 3ms and 3fs object pronominal suffixes and also very occasionally to 2ms and 2fs object suffixes. This construction can occur with all persons of the verb although in the case of the 2fs, 3mp and 2mp, the optional final -n of the verb is never found ; this contrasts with 1s suffix -nī which is occasionally added without loss of the final n.20
2.2.4 It has long been recognised that the contexts in which verb-final n and pronouns incorporating n occur are generally limited to those listed under 1 and 2 for the Arabic muḍāri‘ forms (§2.1.2) ; they do not occur in Hebrew contexts 1 to 4 above where clear examples of GPA forms are found. Both types of GPE form are traditionally explained as expressing emphasis and, in the case of the object-pronominal suffixes, as being related in some way to the Arabic energic forms (particularly the shorter version) where final n is retained before object suffixes.21 This conjecture is supported by the el-Amarna texts where a number of possible energic singular forms occur with final –una (mostly Akkadian forms with apparently Canaanite suffixed morphemes).22 In these forms also the n appears to be retained when followed by an object pronoun, for example uš-ši-ru-na-ši ‘I will send it’, with Akkadian object pronoun –ši.23
2.2.5 If morpheme n in the Hebrew object pronominal suffixes is indeed a fossil remnant of the final element in an original GPE form it follows that many if not all Hebrew GP forms having object pronouns without n are probably GPA. This conjecture is supported by verbs on weak roots, where apparent apocopate forms always occur with the shorter object pronoun and pronouns with n only occur with ‘extended’ verb forms, where the weak radical is retained. Whatever the correct explanation, these pronominal suffixes must either be grammatically conditioned, which although synchronically correct is diachronically very odd or, what seems more likely, the n was an original part of the verb form which become detached when, for whatever reason, in most contexts it became redundant as a grammatical marker.
2.3 Ugaritic
2.3.1 Being closely related to the Canaanite dialects, it might be anticipated that any Ugaritic evidence for hypothesis A would be similar in nature to that of Hebrew, particularly with regard to the distribution of III-weak forms and forms with a suffixed morpheme in n.24 As will be seen, this is to some extent the case, but of all Semitic languages the Ugaritic data is perhaps the most difficult to interpret. This is partly because of the orthography which, aside from the general absence of vocalisation, appears not always to be consistent, and partly due to the poetical nature of the major texts.25 Furthermore the metrical structure of Ugaritic poetry, with its preference for balanced hemistichoi exhibiting approximately the same number and length of syllables, not infrequently appears to result in different forms occurring in what is essentially the same context. Finally it must be suspected that, on occasion, a given form results from the application of particular rules of Ugaritic syntax, giving constructions which at first sight could be taken to refute the hypothesis under examination.26
2.3.2 In principle, the clearest evidence for Ugaritic GPA forms should be provided by forms on III-weak roots in environments where, as in Hebrew, orthography omits the final radical.27 Examination of the relevant forms suggests that they tend to occur in at least three of the four types of construction listed for Hebrew GPA forms in §2.2 above, the exception being that there appears to be no example of a prefixing III-weak root in a conditional or other hypothetical construction.28 Thus, in parallel with the Hebrew waw-consecutive construction there are many Ugaritic examples of type:
wt‘n btlt ‘nt ‘the lady Anat spoke’ (51-3.32)29
where t‘n is 3fs from III-weak root ‘ny ‘speak’. Inspection of the concordance30 demonstrates the formulaic nature of this type of phrase (commonly with 3ms form wy‘n) and that most examples utilise an orthographic GPA form. Examples in cohortative and negative-imperative contexts are:
npš yḥ dnìl ‘with breath let Dn’l be quickened’ (2AQHT-1.37)
bn àl tbkn… ‘my son, do not weep for me…’ (125-25)
The roots here are respectively ḥwy (Hebrew ḥyy31) and bky. Thus, with the exception of conditiomal constructions the Ugaritic examples reflect an original aspect element <singulative> to much the same extent as equivalent examples in Hebrew.32
2.3.3 Although GPA forms tend to be the norm in such contexts, a difficulty arises in the not-infrequent cases where an orthographic GPE form is apparently used to express <singulative> aspect, as for example:
wy‘ny krt tà ‘Krt the noble spoke’ (125-24)
GPE forms on root ‘ny are more the exception and, although in a poetical context a form requiring an English present tense cannot be entirely ruled out, it seems more likely that y‘ny is merely an orthographic variant of y‘n, both perhaps being vocalised *ya‘ni. But where fewer forms of a verb are attested the situation is less clear. For example the concordance lists eight GP forms on III-weak root gly ‘penetrate’, six of which occur in phrases similar to:
tgly dd ìl wtbù qrš mlk àb šnm ‘she penetrated/penetrates El’s field and entered/enters the pavilion of king Abi Shunem’ (51- 4.23, ANET, p133).
where tgly is an orthographic GPE form apparently expressing <singulative> aspect.33 However the seventh occurrence (‘NT.VI-5.15) has GPA form tgl in the identical phrase suggesting that, if tgl is not an error, all seven forms could in fact be GPA, perhaps vocalised *yigli or *tigli. On the other hand there are also cases where an apparent GPE form may in fact be a dual GPA form, a possible example being:
àkr ymġy ktr wkss ... ‘then came Ktr and Kss ... (2AQHT-5.28)
where ymġy should perhaps be vocalised *yimġiyā.34
2.3.4 For evidence of orthographic GPE forms expressing <non-singulative> aspect, first consider forms without a post-root morpheme in n, and in particular forms on III-weak and III-aleph roots. As noted above, some III-weak forms orthographically GPE occur in contexts where a form expressing <singulative> aspect would be predicted, but there are many more examples which are clearly analysable as <non-singulative> and thus parallel equivalent III-weak forms in Hebrew (and Arabic). For example, on roots bky ‘weep’ and mġy ‘reach’ :
àbky wàqbrnh... ‘I’ll weep and inter him…’ (1AQHT-3.140)
rìšh lymġy àpsh ‘his head does not reach its (the throne) top’ (49-1.31)
These forms should perhaps be vocalised *’abkiyu and *lā yimġiyu and can be contrasted with:
ybk làqht ġzr ‘he wept for the young man Aqht’ (1AQHT-4.173)
tmġ ln'm[y arṣ] dbr ‘she came to the goodness of the land of Dbr’ (67-6.28)
perhaps vocalised *yibki and *timġi.35 Orthographic GPA forms rarely occur where a GPE form might be predicted, although a possible exception is:
nġẓ qnyt ìlm ‘we entreat the creatress of the gods’ (51-3.34)
although nġẓ (root ġẓy) would fairly comfortably admit the jussive sense ‘let us entreat’.
2.3.5 The feature of Ugaritic orthography whereby aleph is written in one of three ways, depending on the value of the accompanying vowel, is potentially instructive. It is usually assumed that the vowel represented by the sign always follows its aleph.36 However in some cases the vowel must precede, as for example in yìsp ‘he gathered’ (root àsp), a GP form where the sequence ’s is almost certainly a consonant cluster. Moreover, if Ugaritic did indeed utilise GPA forms the final phoneme in a 3ms form on a III-aleph root must (by definition) have been the aleph, in which case the vowel represented by the sign must again precede its consonant, e.g.:
yqrà mt bnpšh ‘Mt called out to himself’ (51-7.47)
where yqrà clearly expresses <singulative> aspect, and morphologically is probably GPA (perhaps vocalised *yiqra’).37 On the other hand there are examples such as:
ymlù lbh bšmkt ‘her heart filled with joy’ (‘NT-2.25)
where the context is <singulative> and where (poetic licence aside) a GPA form might be expected. But neither Arabic, Hebrew nor Akkadian supports u as the theme vowel of root ml’ and thus on present evidence this and similar forms can only be explained by assuming that the vowel is final and that the forms are GPE.38
2.3.6 Ugaritc verb forms with final n are of three types:
- Singular forms lacking a pronominal suffix;
- Forms with a pronominal suffix;
- Plural and dual forms.
Singular forms without a pronominal suffix are those classified as ‘energic’ by Sivan and Gordon.39 But as with the el-Amarna letters, it is a triumph of the imagination to detect any energic sense in most of these forms,40 particularly for example, in:
àtbn ànk wànkn ‘I will sit and rest’ (2AQHT-2.12)
where both forms are on weak roots (ytb and nwk) and it seems more likely that GPE marker –n is added to give more syllabic ‘substance’. On the other hand, in:
bkm tmdln ’r bkm tṣmd pḥl ‘weeping she saddles a donkey, weeping she harnesses a mule (1AQHT-2.57)
tmdln displays the –n but tṣmd does not. The explanation may be metrical, tṣmd perhaps being a D-form.
2.3.7 Although pronominal suffixes of form –nn suggest that <non-singulative> -n was followed by a vowel, as for example tštnn ‘she sets him’ (1AQHT-2.59), there is almost no evidence for its value, although a possible exception is :
ìqràn ìlm nàmm ‘I shall invoke / let me invoke the kindly gods’ (52. 23)
where Gordon and Sivan take ìqràn to be cohortative (perhaps *’iqra’ana). But whether the à here marks the preceding or following vowel (or both) is unclear; compare the parallel form ìqrà (52. 1).41
2.3.8 Many forms with n occur in conjunction with a 3s suffix in h, although rarely with any other, a situation which again parallels that in Hebrew. Examples are:42
àbky wàqbrnh ‘I’ll weep and inter him ... ’ (1AQHT-3.140)
ḥwy y‘šr wyšqynh ‘...gives a feast for the life-given and bids him drink’ (2AQHT-6.31)
As noted above 3s suffixes also occur as –nn, or simply –n, as:
tštnn lbmt ‘r ‘she places him on the donkey’ (1AQHT-2.57)
àštn bkrt ìlm àrṣ ‘I’ll place it in the [hollow] of the earth gods’ (1AQHT-3.140)
These may correspond to Hebrew suffix forms in –ennu.43
2.3.9 The third type of form with final –n comprises plural and dual forms which, in the former case at least, resemble Hebrew pausal forms. Examples are:
‘l bt àbh nšrm trkpn ‘above her father’s house vultures hover’ (1AQHT-1.32)
tntkn ùdm ‘th km tqlm àrṣh ‘his tears pour forth like shekels to the ground’ (KRT-28).
tšàn ghm wtṣḥn ‘they (dual) raise their voices and cry’ (KRT-267)
Only the first of these is unambiguously <non-singulative>. Possible vocalisations are *tirkapūna, *tunattikūna (compare ytk in 1AQHT-2.82 which omits the first radical n), *tišša’āni and *tiṣaḥāni.44
2.3.10 If it is generally the case that a 3s suffix in -nh signals the presence of a GPE form then it may be that verb forms which have only h as their suffix are GPA, as for example:
ìmkṣh kd ‘l qšth ‘I smote him (but for) his bow’ (1AQHT-1.14)
But the number of forms having only h as a 3s suffix is surprisingly low, although most are aspectually <singulative>. One obvious exception is:
ìbġyh btk ġry ìl spn ‘I will reveal it in the midst of my mountain Godly Zaphon’ (‘NT-3.26)
where ìbġy is morphologically GPE and, in having apparently future reference, is aspectually <non-singulative>.
2.3.11 To summarise, where morphologically GPA and GPE forms can be identified with reasonable confidence, the hypotheses ‘GPA ⇔ <singulative>’ and ‘GPE ⇔ <non-singulative>’ are supported, but not without exception. There is a considerable grey area, relating particularly to III-weak and III-aleph forms, where verbs which appear to be morphologically GPE are used in both <singulative> and <non-singulative> contexts. Similarly, although most forms incorporating final or near-final n occur in <non-singulative> contexts, there are some whose context is <singulative> and would on occasion be better translated by a past tense. But although such data could be viewed as at least a partial refutation of Hypothesis A (along with the not-insignificant number of GS forms apparently expressing <singulative> aspect – see Section 7) would it be reasonable to expect verbs in poetic and literary texts always to exhibit a rigid correlation with their postulated aspectual functions? Moreover, Gordon has drawn attention to cases where a text appears to consist of an assemblage of fixed phrases and, if this were a productive technique in Ugaritic poetry, it could account for at least some forms which are otherwise difficult to analyse.
2.3.12 ‘Scholarly’ analysis of the Ugaritic verb tends to be rather traditional. Sivan, for example, divides prefixing verb forms into ‘indicative and ‘injunctive’, each category consisting of three subdivisions.45 Thus for G-forms:
Table 2.1 Ugaritic Verb Types
Indicative | Injunctive | ||
---|---|---|---|
Preterite | GPA | Jussive | GPA |
Imperfect | GPE(u) | Volitive | GPE(a) |
Energic | GPE(un) | Energic | GPE(an) |
where the letters in brackets distinguish GPE form variants. There are several problems which considerably weaken this formulation as a possible refutation of Hypothesis A :
- Sivan assumes that jussive and preterite are not only synchronically but also diachronically distinct forms. This analysis is presumably founded on the differing stress patterns marked by the Masoretic vocalisation of equivalent Biblical Hebrew forms (see footnote 18 from §2.2.1), and the fact that their synchronic grammatical functions appear irreconcilable. But it cannot be shown that in Ugaritic (or for that matter in Akkadian or other languages) preterite and jussive forms actually exhibited differing stress patterns nor that the differing patterns in Hebrew are not secondary.46 This supposed grammatical irreconcilability can of course accounted for by the proposed aspect element <singulative>.
- As has already been noted, the majority of supposedly ‘energic’ forms contain no such implication,47 so that the indicative imperfect and energic forms can readily be collapsed into a single sub-division where the ending can be either –u or –un for singular and 1p forms (excluding 2fs), depending on context, metre or the presence or absence of a pronominal suffix.
- The distinction between indicative and injunctive energic forms seems to have been drawn mainly for purposes of symmetry ; there is little evidence for it in the data.48 Almost the only potential evidence is the phrase ìqràn ìlm n‘mm (52-23), translated ‘let me invoke the goodly gods’ by Sivan and ‘I invoke the good gods’ by Gordon.49 This phrase should be compared with the almost identical ìqrà ìlm n‘mm (52-1). Gordon’s translation seems to imply the vocalisations *’iqra’un and *’iqra’u and Sivan’s the vocalisations *’iqra’an and *’iqra’. It seems impossible to tell which of these, if either, is correct. As there is little other evidence for the energic injunctive then, if Gordon is correct, the case for such a form is almost completely undermined.
- Although there is no reason in principle to doubt the existence of a ‘volitive’ form in Ugaritic, perhaps as a secondary formation like the Arabic subjunctive, Sivan’s evidence is again drawn entirely from verbs on III-aleph roots, and thus rests to a considerable extent on the claim that the vowel always follows its aleph. In consequence he analyses as volitive, forms which in all probability are declarative GPA, as for example wymẓà (75-1.37). This verb is preceded by two which appear to be declarative (yḥ and ymġy, probably vocalised *yiḥi and *yimġiyū50), and even on Sivan’s translation (which assumes that Ba‘al is the subject) wymẓà is declarative, and therefore should perhaps be vocalised *yimẓa’.51
- In sum, there is no evidence that preterite and jussive forms were morphologically or phonologically distinct in Ugaritic, so that the two can readily be collapsed into the single category ‘GPA ⇔ <singulative>’. Likewise there is little convincing evidence for Sivan’s volitive or for his energic injunctive, which latter should almost certainly be collapsed with the equivalent indicative forms. But nor is there good evidence for distinct indicative imperfect and energic forms, so that these four categories can without serious difficulty be collapsed into the single category ‘GPE ⇔ <non-singulative>’ proposed in this study.
2.4 Old Aramaic
2.4.1 Of the range of functions for Biblical Hebrew GPA forms set out in §2.2.1, the waw-consecutive construction in Old Aramaic52 (OA) is attested only in w’s’ ydy wy‘nny ‘and I raised my hands and he answered me’ in the Zakir inscription. Segert takes this to be a Canaanism (AAG, §6.6.3.3.2, p377), but the statement is formulaic and could well reflect an earlier stage of the language where such constructions were more common.53 The form wy‘n-ny (root ‘ny) is orthographically GPA, although given the tendency in early Aramaic towards abbreviated orthography it does not necessarily follow that this is indeed a GPA form.54
2.4.2 In contrast to Arabic and Hebrew there are in general relatively few morphologically apocopate II-weak and III-weak GP forms in OA (AAG, §5.7.6.4.5), the majority of such verbs being morphologically GPE. But as in Hebrew, an -n suffixed to the 2fs, 3mp and 2mp persons of Aramaic GP forms marks the relevant forms as GPE,55 so that GPA forms only become apparent for verbs in ‘jussive’ constructions with a plural or feminine singular subject, where the final n is omitted.56
2.4.3 According to the Masoretic pointing, object pronominal suffixes in Biblical Aramaic (BA) are almost always prefixed by inn, nn or n. These occur typically in an ‘imperfect’ context, the only (rare) examples without n occurring in a ‘jussive’ context57 ; this statement appears also to be true for OA generally and is reminiscent of the situation in Ugaritic (§2.3.9 above).58 The Aramaic suffixes thus syntactically resemble those of Hebrew and Ugaritic and the associated verb forms can probably be similarly construed as GPE and GPA respectively. However, insofar as the data permits judgment, the object pronouns of OA differ generally from those of Hebrew and Ugaritic in having the n(n) element whatever the number and person of the suffix. As in Hebrew, the final n of the 2fs, 3mp and 2mpGPforms is omitted before suffixed pronoun n(n).
2.4.4 According to Segert (AAG §7.5.4.1.1) ‘jussive’ (GPA) forms can be used in the apodosis of conditional clauses, but no example is cited so it is not possible to tell whether these relate to ‘possible’ or ‘impossible’ conditions.59
2.4.5 To summarise, morphologically and functionally, Aramaic GP forms can without difficulty be correlated with those of Hebrew and Arabic and therefore support the hypothesis under investigation more or less to the same extent as the data in the latter two languages. In general Aramaic verb forms stand closer to those of Arabic than Biblical Hebrew, in that the primary contrast is between a GS form typically signalling an event in past time and a GP form signalling events in all other time and aspect frames, except those expressed through participles. However, the repertoire of GPE forms available to Arabic is not present in Aramaic, either because in all likelihood they never existed (e.g. the subjunctive form) or because they have for the most part coalesced to give a GP paradigm comprising both GPA and GPE forms.
2.5 Akkadian
2.5.1 The obvious example of a GPA form in Akkadian is the ‘preterite’, built on templates iqbar, iqbir and iqbur.60 These forms are typically used in the following contexts:
- To express <singulative> events occurring in past time (GAG §79);
- Non-declarative constructions such as. jussive and cohortative (GAG §81);
- Negative imperatives (GAG §81);
- Conditional clauses (GAG §161).61
In the first of these the use of the preterite resembles that of GPA forms in Hebrew waw-consecutive constructions and also, albeit less clearly, GPA forms in Ugaritic. The jussive, cohortative and negative imperative functions also parallel those of Hebrew and Arabic, while the fourth function parallels the use of the GPA form in certain types of Arabic and Hebrew conditional clause. Thus none of these functions presents any new problem for hypothesis ‘GPA ⇔ <singulative>’.
2.5.2 Akkadian GPE forms are the ‘subjunctive’ and ‘ventive’, which in the G-stem are typically formed respectively on templates iqburu and iqburam.62 With its essentially directional sense the ventive is taken to be an Akkadian innovation, although note that in later texts the final -m tends to be omitted, giving a form morphologically reminiscent of the Arabic subjunctive. Subjunctive forms in the later Assyrian dialects are further characterised by lengthening of the u and the addition of final morpheme ni 63; in Old Assyrian –ni is restricted to the 2p and 3p forms. Thus in certain respects the Akkadian GPE subjunctive resembles GPE forms in Arabic, Hebrew and Aramaic. But syntactically, the Akkadian form is confined to subordinate clauses - relative clauses in particular, typically introduced by relative pronoun ša. The Akkadian subjunctive is also formed on the permansive and the present, in additon to the preterite64 and thus, although the absence of evidence from other languages suggests that certain of these forms may be Akkadian innovations, the preterite subjunctive offers ambigous support for hypothesis ‘GPE ⇔ <non-singulative>’.
2.5.3 GPA form iqbur contrasts primarily with ‘permansive’ (GS) qabir and ‘present’ iqabbar, the functions of the latter resembling those of the GPE forms of Arabic, Hebrew, Aramaic and Ugaritic.65 This form cannot be explained by Hypothesis A and is discussed under Hypothesis B.
2.6 Ge‛ez
2.6.1 GPA forms in Ge‛ez are confined to the ‘subjunctive’, conjugated on templates yəqber (transitive) and yəqbar (intransitive). These forms have jussive, cohortative and negative-imperative function, in common with the Arabic majzūm, but serve in addition to express purpose or result.66 As with Arabic, none of these functions necessarily implies iteration or duration, so that the subjunctive form presents no serious problem for hypothesis ‘GPA ⇔ <singulative>’. In expressing purpose or result the Ge‛ez form is of course reminiscent of the Arabic subjunctive and, given that the Ge‛ez form is clearly GPA, this functional match offers support for the conjecture, outlined in §2.1 above, that the Arabic subjunctive is a development of the GPA form (majzūm) internal to Arabic.67
2.6.2 Ge‛ez exhibits no form interpretable as GPE and certainly none with a final n-based morpheme. The Ge‛ez form used to express non-singulative senses has the structure yəqabbər, which both morphologically and functionally resembles Akkadian iparras and is therefore likewise discussed under Hypothesis B.
2.7 Epigraphic South Arabian
2.7.1 Being preserved in an unvocalised script, the Epigraphic South Arabian (ESA) evidence poses similar problems to those of Ugaritic. As with Ugaritic - and Hebrew - the most promising source of evidence for GPA forms should in principle be verbs on III-weak roots. However there appear to be relatively few of these, although one clear example is:
ly’t ‘dy ’twtm ‘may there come from Itwat’.68
This construction is of course reminiscent of the ‘optative’ use of the Arabic GPA form, a possible link that must be treated with caution since such forms, again as in Ugaritic, may be graphically rather than morphologically inspired.69
2.7.2 Evidence for GPE forms in the ESA dialects is provided by the common Sabaic phenomenon of verbs with one or more final n. According to Nebes and Stein, singular verbs of this type are formed on the pattern yqtln (3ms) and plural forms on yqtlnn (3md, 3mp).70 The former could without difficulty be vocalised *yiqtulun but the latter is more problematic. But Höfner’s discussion, and the comments of Nebes and Stein, indicate that -nn forms are almost entirely confined to Sabaic and thus may be a secondary innovation within that dialect ; however, Minaic 3fp form tqtlnn is also cited, although compare 3mp form yqtlwn in Qatabanic.71
2.7.3 The ESA script does not mark gemination, so that sequence –nn must represent distinct morphemes. The -w of GPA 3mp form yqtlw probably signals final –ū rather than a diphthong, and it is also likely that in the Qatabanian GPE form yqtlwn final –wn is the standard Semitic masculine plural morpheme –ūn. Thus it is posssible that the latter morpheme is also represented by the first n in Sabaic sequence -nn. The function of the second –n then remains obscure, but could it be that the proposed GPE singular marker *-un has been replicated in the plural forms by analogy? This would suggest a vocalisation *yiqtulūnun.72
2.7.4 Nebes and Stein state that over 75 per cent of the attested Sabaic prefixing verbs forms are of type yqtln, but also observe that ‘in Early Sabaic (yqtl) is attested considerably more often than (yqtln)’, although they do not indicate whether these patterns occur in similar or different contexts.73 They also state: ‘the long form of the imperfect, seldom the short form, occurs in statements with present and future reference’, thus supporting, as far as it goes, hypothesis “GPE ⇔ <non-singulative>”. But as with Ugaritic there are difficulties, as for example in:74
fygb’w ‘dy hgrn n‘ḍ wbnhw fyhṣrn mlkn ’ls2rh yhdb wdbn kms1hw ... wyqm‘w whb‘ln hgrnhn ‘from there they went to the city of N‘ḍ [and its environs]. Thence king I-Y along with part of his army set out for ... Then they destroyed and seized both cities’.
There can be little doubt that these events are aspectually <singulative>. The first and fourth verbs (fygb’w and wyqm‘w) are GPA and as such support “GPA ⇔ <singulative>”. However although apparently in the same <singulative> context, fyhṣrn is GPE, which is difficult to explain unless the author wished to suggest that ‘set out for’ was a more protracted process requiring a <non-singulative> form. The fourth form whb‘ln is infinitive and used in this way can be understood as an ESA innovation (Beeston, DGESA, §19-5).
2.7.5 Hypothesis “GPA ⇔ <singulative>” is also supported by jussive and negative imperative constructions, as for example:
wl ydbḥw bn ms2mnhm ‘and let them offer (an animal sacrifice) from both culitvated areas’;
w’l yhwfd bh ‘md w ‘lbm ‘and neither vines nor ‘lb shall be planted there’. 75
where yhwfd is a (passive?) S-form. On the other hand ESA clauses indicating possible conditions not uncommonly incorporate a GPE form in final –n, and obviously as such do not support ‘GPA ⇔ <singulative>’.76 Such constructions are of course reminiscent of Arabic conditional clauses incorporating an energic GPE form (§2.1.3 above) and it may be either that both represent weakening or modification of the original use of a GPA forms in such constructions or, more troubling for Hypothesis A, that the Arabic and ESA forms share a common origin.
2.8 Modern South Arabian
2.8.1 The Modern South Arabian (MSA) dialects display a number of GP forms. Taking Mehri as an example, there are three GPA forms associated with regular verbs, namely : an active subjunctive (3ms yərkēz), a passive/imperfect subjunctive (yərkōz) and an intransitive subjunctive/imperfect (yətbōr). The active subjunctive is paralleled by an imperfect form on pattern GVP (3ms yərūkəz).77 All three types then have a GPE ‘conditional’ form, active yərkēzən, passive yərkīzən and stative yətbīrən.78 As in Arabic (§2.2.1) the passive forms are taken to be a later innovation, a position supported by the fact that in most respects the passive conjugations parallel those of the presumably older intransitives.
2.8.2 Both in form and function the Mehri subjunctives are reminiscent of the equivalent Ge‛ez GPA forms79 and support or refute Hypothesis A to the same degree as the latter. In having final -ən the Mehri conditional forms are self-evidently of type GPE. But, equivalent forms being absent from the other MSA dialects together with the fact that their use in Mehri is confined to the apodosis of ‘impossible’ conditional clauses (TSM §2.5.1.3.2.3), would suggest that they cannot be adequately explained by “GPE ⇔ <non-singulative>”.80
2.8.3 The diachrony of the GVP active imperfect form (3ms yərūkəz) poses a particular problem. There are two competing hypotheses ; one that it originates in a form similar to Ge‛ez yənaggər or Akkadian iparras (see the following section), and the other that it represents a modification of a Semitic *yiqburu(n) that has lost its final n. Cohen, for example, proposes for Śḥeri a sequence through which an original *yiktubu could have evolved into the attested ikóteb (ESVS p73).
2.8.4 Others however argue (for example Lipiński, OCG, §38.5) that rather than originating in *yarkázu or similar, Mehri yərūkəz originates in *yarákkaz or similar. But Cohen points out (ESVS p75) that stressed vowels in Mehri are always long, whether in closed or open syllables, so that the long second syllable is more likely to be a secondary consequence of its taking the accent rather than arising through loss of a geminate consonant. Support for Cohen’s interpretation is offered by Mehri forms on geminate and II-weak roots, which are difficult to explain on the basis of original forms equivalent to yənaggər. For example geminate 3ms imperfect yədlūl (subjunctive yədēl) has no vowel between first and second radical (cf. Ge‛ez yədannən), and in II-weak imperfect yəmūt (subjunctive yəmēt) the weak consonant is not preserved (cf. Ge‛ez yəmawwət).
3. Evidence for Hypothesis B
3.1 The second possibility envisaged for expressing element <non-singulative>, alternative to morphemes suffixed to the triradical stem, is gemination of the second root morpheme (§1.5 above). In the Semitic languages there are three important prefixing forms with geminate second radical (type GPG), namely ; the Ge‛ez and North Ethiosemitic ‘imperfect’ form yənaggər, the iparras form of Akkadian and the D forms (and their derivates) in the various languages.81 In the N.W. Semitic and S. Ethiosemitic languages forms similar to yənaggər and iparras are attested only spasmodically, if at all.82
3.2 The functions of Ge‛ez yənaggər are similar to those of the Arabic muḍāri‘ (for example) and to that extent support the “GPG ⇔ <non-singulative>” strand of Hypothesis B. But the evidence of the MSA languages seems crucial here for if, as suggested at §2.8.3, the Ge‛ez and MSA subjunctives share a common origin, the same may well be true of the associated ‘imperfect’ forms. Thus if MSA imperfects originate in GPE yaktubu(n) or similar (§2.8.3), the same could well be true also of the Ge‛ez imperfects, the latter having perhaps evolved along the lines yaqtúlu → yəqātel → yəqáttəl. In other words, despite assertions that N. Ethiosemitic imperfectives derive from a Common Semitic original with gemination of the second radical, they could perhaps better be viewed as a secondary and (relatively) late formation which should not be adduced as evidence for Hypothesis B.83
3.3 Akkadian iparras is also functionally reminiscent of the Arabic and Hebrew GPE forms and hence could in principle be explained in the same way as proposed for Ge‛ez yənaggər. But iparras provides better evidence for Hypothesis B, in part because Akkadian GPG forms on occasion display the same theme vowel as their associated GPA forms, which is perhaps less likely to have occurred if iparras were a secondary formation.84 But importantly, Akkadian subjunctive iqburu - a GPE form - did not also evolve into a GPG form, as might have been anticipated had iparras also originated in a GPE form. The Akkadian GPG form thus seems more likely to be ‘primary’, rather than ‘secondary’ as proposed for the Ge‛ez form. But if this was the case it would appear to follow that GPE and GPG forms must at one time have co-existed in Common Semitic but that the latter were for the most part lost from languages other than Akkadian.
3.4 But this conclusion begs the question as to why two separate forms (GPE and GPG) should apparently have been used to express the same aspect element. However, note firstly that although element <non-singulative> can encode the semantic element [iterative], this sense of repetition could originally have been captured by exploiting the onomatopoeia of a reduplicating (pre-Semitic) biradical stem to give a prefixing (GPR) verb of the form *ipurpur. Then, if the stress patterning rules governing the triradicalisation of an original biconsonantal stem and augment are applied to a reduplicated biradical stem and its augment (refer to §8.6 in MPSVS), it can be argued that the resulting triradical form tends to display gemination in its second radical.
3.5 It is therefore proposed that iparras evolved from a reduplicating biradical form *ipurpur plus augment, expressing an aspect element <iterative>, alongside a GPE form expressing the other semantic components of <non-singulative>. As part of this evolution iparras is taken to have assimilated all (proto-) Akkadian reduplicating (GPR) forms, for it is striking that Akkadian seems to be entirely lacking in reduplicated roots (GAG §108-10), in contrast to the other Semitic languages, where a number of unaugmented reduplicates appear to have been preserved more or less in their original form85. But this begs the question as to why Akkadian did not also preserve at least some such forms, especially as they are common enough in Sumerian, with which Akkadian was in close contact.86
3.6 Further support for a GPG form among the original repertoire of Semitic verbs is the widespread incidence of D forms in the various languages. By contrast, derived verbs in the Cushitic, Omotic and Chadic languages include no form equivalent to the D form (apart from loans) and neither apparently does Egyptian, suggesting that the D form could be seen a Semitic innovation, for which the likliest source would have been a GPG form. But the differing vocalisations of the Akkadian GPG and DP forms is an obvious obstacle to such a proposal, such that if the latter indeed derives from the former the GPG form must in certain contexts have adopted the vocalisation, say, of the SP-form (cf. Akkadian ušapris vs uparris) ; such an account would of course require the associated DPA and (suffixing) DS forms to be subsequent innovations.
4. Hypothesis A Refuted?
4.1 The discussion in Section 3 suggests that, at least in the case of Akkadian iparras, Hypothesis B cannot be entirely refuted, but if so what then is the status of Hypothesis A? This section summarises and reviews the evidence.
4.1 GPA ⇔ <Singulative>
4.1.1 To recapitulate ; <singulative> aspect is defined as encoding an event which the user views as having occurred only once, or whose interest in an event that has not yet taken place is confined to a single potential occurrence. Whether the event has actually occurred or will occur more than once is taken not to be relevant (§1.2 above). <Singulative> aspect is then postulated to have been expressed by a GPA-type verb of typical form (3fs) *tinpur (§1.5). Dividing the GPA forms discussed in Section 2 into those whose functions are ‘declarative’, signalling an event which has definitely occurred, and those which are ‘non-declarative’, Table 4.1 summarises their distribution among the various languages. It will be seen that whereas jussive/cohortative and negative imperative functions are attested for GPA forms in all the languages, declarative and conditional functions are rather less prominent.
Table 4.1 Distribution of Proposed GPA Functions
Language | Declarative | Non-declarative | ||
---|---|---|---|---|
Conditional (Possible) |
Jussive / Cohortative |
Negative Imperative |
||
Arabic | X | X | X | |
Hebrew | X | X | X | X |
Ugaritic | X | ? | X | X |
Aramaic | X | X | ||
Akkadian | X | X | X | X |
Ge‛ez | X | X | ||
MSA | X | X | ||
ESA | X | X | X |
4.1.2 In those languages where declarative <singulative> events are not expressed by a GPA form a GS form is utilised (Section 7). But despite the somewhat inconclusive nature of the evidence for the declarative and conditional functions of the GPA form, and notwithstanding the fact that declarative and non-declarative functions are logically distinct, the likliest explanation for the phenomena summarised in Table 4.1 remains that they do indeed have their origin in a Common Semitic aspect element <singulative> and that hypothesis “GPA ⇔ <singulative>” is the best morphological fit for the evidence.
4.2 GPE ⇔ <Non-singulative>
4.2.1 <Non-singulative> aspect is defined as encoding all events which the user cannot or does not wish to encode as <singulative> and is postulated to have been expressed by a GPE form (§1.2 and §1.5 above).
4.2.2 The conventional explanation for Hebrew, Ugaritic and ESA forms incorporating final n or nn, in the ways outlined in Section 2, is that they are in some way related to Arabic energic forms and that the absence of energic sense results from weakening of the original function.87 But consideration of the data from first principles has suggested that something like the reverse may originally have been the case, namely :
- Suffixed n was integral to the original GPE form, and formed part of the morpheme(s) marking <non-singulative> aspect;
- Energic or pausal functions inferred from context originate in those of the original GPE form and are diachronically secondary.
Only in Arabic is the energic connotation clearly dominant in GPE forms with final or near final n. There may be energic forms in the el-Amarna texts but ‘there is no way to distinguish in the script between the indicative imperfect (GPE) plural and the indicative energic plural’.88 Thus in both Arabic and el-Amarna it is arguable that the n-based suffixed morpheme of energic forms (actual and putative) originates in the proposed marker of <non-singulative> aspect.89
4.2.3 In ESA the syntactic distribution of GPE forms tends to be wider than in the other languages. Höfner partly explains this by assuming that the functions of the n-form were extended to replace a ‘subjunctive’ which had lost its distinctive morphological structure.90 But of course this explanation, and indeed the problem it seeks to answer (the absence of a ‘subjunctive’ form), assumes that the situation in Arabic most nearly reflects the original verb morphology of Semitic. But if, as has been suggested above and is further discussed below, the Arabic subjunctive is a development internal to that language, the distribution of the ESA forms in final n would be more elegantly explained by proposing that this morpheme was originally part of the of the GPE form used to express <non-singulative> aspect, and that its use was subsequently expanded to express such connotations as ‘subjunctive’.
4.2.4 But the ESA GPE form also occurs in the apodosis of ‘possible’ conditional clauses, which at first sight refutes one element of Hypothesis A, but on the other hand could simply (although not entirely convincingly) be explained as an extension of the use of the GPE forms91 The Mehri of Oman also displays ‘conditional’ forms in final -ən, employed in the apodosis of ‘impossible’ conditionals (Watson TSM §2.5.1.3.2.3), but as similar forms do not occur in any other MSA language it may well be an innovation in the Omani dialect ; compare also the use of the Arabic energic in the apodosis of ‘possible’ clauses (Wright, GAL, Vol II §19c).
4.2.5 The 3ms GPE form in Common Semitic may therefore have been of the general form *yinp(v1)r-(v2)n.92 With regard to the value of v2, in Arabic and el-Amarna the vowel following the final radical in the 3s, 2ms, 1s and 1p forms of the GPE form is u, whereas in the equivalent Arabic energic forms the vowel preceding the n is a.93 The final vowel of the Akkadian ‘subjunctive’ is also u although, as noted at §2.5, the diachronic relationship between this form and GPE forms in the other languages is not clear. The vowel preceding Biblical Hebrew and Biblical Aramaic n-type pronominal suffixes is that of the later Masoretic vocalisation. Thus, with the partial exception of the Arabic energic, in those cases where the vocalisation is contemporary with the documents themselves the value of v2 is u. Therefore, if it is correct to understand the morphemes marking the Arabic energic forms as deriving from a suffixed morpheme expressing <non-singulative> aspect in Common Semitic, the post-radical vowel in the relevant forms may originally have been u and subsequently shifted to a.
4.2.6 As to plural GPE forms, the hypothesis that –ū was the original 3mp/2mp marker in Semitic is strongly supported in the various languages,94 so that the Common Semitic 3mp forms may have been (GPA) *yinp(v)rū vs (GPE) *yinp(v)rūn. The latter form does not fit easily with the Arabic energic forms but a possible explanation may be sought in the conjecture that a long vowel may on occasion be replaced by gemination (§3.2 above), so that the proposed form *yinp(v)rūn could have become *yinp(v)runn(a). But it would then have to be assumed that the longer Arabic 3ms energic form yaqtulanna and its equivalents were formed by analogy with the 3mp/2mp forms and that the shorter 3mp energic form yaqtulun evolved by analogy with the shorter 3ms form and its equivalents.
4.3 Hypothesis C
4.3.1 On the basis of the discussion in Section 3 it is likely that Hypothesis B is at least partly valid and as such constitutes a partal refutation of Hypothesis A, in that the definition of <non-singulative> may be too general and need to be modified, at least to exclude events codified as <iterative>. But Hypothesis B alone cannot adequately explain the phenomena discussed under Hypothesis A and it thus seems inevitable that a further hypothesis, Hypothesis C, should be proposed to embrace the GPA, GPE and GPG forms discussed above, namely :
GPA ⇔ <singulative>
GPE ⇔ <non-singulative>
GPG ⇔ <iterative>
4.3.2 But in proposing Hypothesis C it is important to note that if ‘iterative’ implications are stripped away from <non-singulative> there is a somewhat diminished content left to this element, unless it be founded on semantic element [durative], as a contrast to [iterative].
5. GP Subjunctive Forms
5.1 As with the energic forms, it is unlikely that a subjunctive form would be proposed for Ugaritic and el-Amarna if it did not exist in Arabic. But even on this foundation the argument is not convincing. For instance, for el-Amarna Rainey is forced to conclude that the ‘texts have not given us any conclusive evidence for the existence of a Canaanite yaqtula pattern’, since ‘the -a suffix almost always was attached to verbs which tend to be employed with the [Akkadian] ventive’.95
5.2 Ugaritic examples cited by Sivan in support of his ‘volitive’ form are of necessity on verbs with III-aleph roots, and his discussion assumes that the vowel signalled by the aleph always follows its consonant.96 But, as is argued above (§2.3.5), unless verbs on such roots never occurred as singular GPA forms there must be contexts in which the vowel precedes its aleph ; thus a 3ms GPA form yqrà on root qr’ could only have been vocalised along the lines yiqra’. Therefore if some of Sivan’s examples can be equally interpreted as expressing <singulative> aspect, the evidence in support of his ‘volitve’ form becomes rather weak.
5.3 Thus, given the weakness of evidence for a subjunctive form other than in Arabic, and ignoring GPA forms in other languages denoted by the same term, the subjunctive seems most likely to have been an Arabic innovation and is therefore taken not to have been part of the repertoire of verb forms in Common Semitic.
6. Aspect in Egyptian Verb Forms
6.1 Introduction
6.1.1 It is generally accepted that there is a relationship between Egyptian and the Semitic languages in addition to their supposedly common ‘Afroasiatic’ heritage closer than, say, between Semitic and Cushitic.97 Thus despite the undoubted problems, not least the limitiations of the largely non-vocalic hieroglyphic script, comparison of the Egyptian and Semitic verbal systems is a potentially instructive method of testing the hypotheses explored in the previous sections.
6.1.2 Egyptian verb forms98 are based predominately d on triradical roots, although many have a first or third radical that can be understood at best as a weak consonant but which may simply have been a vowel, at least in the earlier stages of the language.99 Egyptian also exhibits a good number of biconsonantal roots/stems, some of which are almost certainly of African (‘Afroasiatic’) origin although others may well be worn-down triconsonantals.100 Nevertheless, comparison with the reconstructed verbal system proposed, say, for Lowland East Cushitic, suggests that the Egyptian system is predominantly Asiatic in nature, and therefore more likely to be related to that of Common Semitic.101
6.2 Outline of Egyptian Verb Morphology
6.2.1 Egyptian displays verb forms analogous to Semitic G-forms, the ‘old perfective’ for example bearing a clear morphological and functional resemblance to the Akkadian ‘permansive’ and therefore, in the notation used in this study, can be classed as GS. Two other forms, which comprise the primary subject matter of this section, are denoted śdm.f and śdm.n.f, where final f is in each case the 3ms subject pronoun.102 In marked contrast to the Semitic languages no Egyptian verb form has subject pronouns prefixed to the stem. All pronominal distinctions are made through suffixed pronouns, so that the śdm.f and śdm.n.f, forms are respectively of morphological type GSA (apocopate) and GSE (extended). The orthography śdm.f appears in fact to embrace at least three different forms, ‘perfective’, ‘imperfective’ and ‘passive’. As is suggested for example by the ‘mutable’ verbs, the orthography śdm.n.f probably also embraces an ‘imperfective’ form, but not apparently a ‘passive’ form.103 A second and more common passive śdm.f form, śdm.tw.f, (RSEVS p318), is taken to be an Egyptian innovation, perhaps originating in the (African) suffixed t-based deriving morpheme, and is not here discussed further.
6.2.2 In the absence of any form resembling a Semitic GP form Egyptian would initially appear to be of little use for testing the validity of the diachrony proposed above for the Common Semitic verbal system. A widely-held opinion is that Egyptian did originally exhibit prefixing verb forms resembling those of Semitic, but that the old system collapsed and was replaced by the śdm.f and śdm.n.f forms.104 Thacker, in contrast, observes that ‘there is not the slightest trace of any such [prefixing] formation in the oldest-known inscriptions ... The only possible conclusion is that Egyptian never at any stage in its history possessed a form with prefixed pronominal elements’.105 Thacker’s position is consistent with a major assumption underlying this study, namely that languages will almost always retain fossilised remnants of earlier structures, the corollary being that if there is no evidence for a structure then it is likely, although not certain, that the structure never existed.
6.2.3 Thacker is further of the opinion that the śdm.f and śdm.n.f forms developed from an early form of the old perfective.106 Gardiner on the other hand argues that they originate in a passive participle, conjugated by means of genitival pronominal suffixes, and in the case of the śdm.n.f form also incorporating a dative n-based morpheme.107 That these pronouns are related to the genitival suffixes is not disputed, but Gardiner argues that ‘only on (the) theory [that the narrative śdm.f and śdm.n.f forms must ... be derived from the passive participles] can the use of the suffix-pronoun in the śdm.f form be explained; it is a direct genitive ...’. 108 But it would seem that the role of the suffixed pronouns in Egyptian became wider than that of their equivalents in Semitic. For, comparison of the Egyptian independent pronouns with equivalent Semitic forms suggests that, excepting the 1s form (Egyptian ìnk, cf. Akkadian anāku) and possibly the 1p form (Egyptian ìnn; Akkadian nīnu), the Egyptian forms have been modified by adding the equivalent suffix pronouns.109 It would seem inescapable that when added to the independent pronouns the suffix pronouns must have lost their original genitival function, suggesting that these forms may have tended to become ‘all purpose’ personal pronouns, except in the case of the singular ‘dependent’ pronouns, where a different series is used.110
6.3 The Functions of the śdm.f and śdm.n.f Forms
6.3.1 In Middle Egyptian the active śdm.f and śdm.n.f forms have a rather complex range of functions. In narrative, the śdm.n.f form generally expresses an event occurring in past time and the śdm.f form events which typically occur in present or future contexts.111 At first sight therefore the śdm.n.f form bears a functional resemblance to the GPA form *yinp(v)r proposed for Common Semitic (§4.1.1 above) and the perfective śdm.f form appears approximately to parallel the proposed Common Semitic GPE form *yinp(v)run (§4.2.3).
6.3.2 But a number of features of the way in which the active śdm.f and śdm.n.f forms are deployed merit further consideration. First of all Gardiner remarks : ‘In Old Egyptian the non-geminating [‘perfective’] śdm.f is fairly frequent in past narration with verbs showing an object, but towards Dynasty VI the śdm.n.f form can be seen gradually superseding it in this use’.112 Secondly, in Middle Egyptian n śdm.f is the ‘common and normal negation of śdm.n.f in past narrative’ and ‘a broad survey shows that the construction n śdm.n.f is common in characterisations, statements of custom and generalisations of all kinds’.113 Taken together, these phenomena suggest that the use of śdm.n.f for past narration may not be original.114
6.3.3 Thus, having noted that the ‘past-narrative’ function of the śdm.n.f form is reminiscent of that proposed for the GPA form in Common Semitic, and that the present/future implications of the perfective śdm.f form resemble those of the GPE form, there is now the suggestion that at an earlier stage something like the reverse may have been the case, a conjecture further supported by the fact that the śdm.f form generally ‘contains no implication of repetition or continuity’,115 irrespective of its time frame, and as such could be interpreted as expressing aspect element <singulative>.116 A further functional similarity between the ‘perfective’ śdm.f form and the proposed Common Semitic GPA form is that the former also is used in the protasis and apodosis of ‘possible’ conditional clauses, the use of the śdm.n.f form in such clauses apparently being restricted to unfulfilled (‘impossible’) conditions, which could be a consequence of its original association with <non-singulative> aspect.117
6.3.4 The discussion in Section 2 above implies that the proposed original function of the GPE form resulted in daughter forms in the Semitic languages coming to express events occurring in present or future time. Although this would appear to conflict with the śdm.n.f form serving to signal events in past time, in many contexts the śdm.n.f form not unexpectedly requires an English perfect tense in translation; in other words such events are not purely ‘singulative’, as defined in Section 1, but imply subsequent ongoing relevance. Thus on the strictest definition of the element <singulative>, śdm.n.f forms requiring the English perfect must formally express <non-singulative> aspect (see Figure 1), and it could be that a further shift may then have taken place whereby the implication of ‘ongoing relevance’ would have been modified, or eliminated, such that the śdm.n.f form came eventually also to express <singulative> aspect.
6.3.5 The morphology of the Egyptian G-forms therefore begins to look rather more interesting. First of all, suppose that the perfective śdm.f form did indeed originally express aspect element <singulative> and that the śdm.n.f form expressed <non-singulative>. The obvious next step is then to note the morpheme represented by n in the śdm.n.f form and to recall that the (3ms) GPE form postulated for Common Semitic is *yinp(v)run, differentiated from the equivalent GPA form by its final n-based morpheme (§4.2 above). Could phoneme n in the proposed Common Semitic GPE form and that of the Egyptian śdm.n.f form be one and the same? This question is further explored in Section 7 of MPSVS.
7. Semitic Suffixing Forms
7.1.1 In the Semitic languages, where declarative <singulative> events are not expressed by a prefixing form a suffixing form is utilised. But in Akkadian and Egyptian, GS forms (‘permansive’ and ‘old perfective’ respectively), are largely confined to expressing events or states not involving an agent. This usage is generally taken to be original, such that Semitic suffixing forms expressing a sense requiring an agent are considered to be diachronically secondary,118 and therefore not to refute the hypotheses under consideration here. This later usage perhaps evolved via an intermediate stage where suffixing forms came to express senses requiring an English perfect tense in translation.
7.1.2 The details of this evolution of course vary from language to language. For example, in Biblical Hebrew, suffixing and prefixing forms requiring an agent occur more or less equally, in rather complex syntactic structures, in a way which hardly occurs in the other languages. The general situation with regard to ‘possible’ conditional clauses utilising suffixing forms is also complex, not least because such clauses are logically ambiguous and so have become an area of considerable fluidity among the languages, a situation which tends to weaken any hypothesis proposed for their history, including the one explored here.119
Bibliography
APPLEYARD, D. L., Beja as a Cushitic Language, in Takács, C. (ed.) Egyptian and Semito-Hamitic (Afro-asiatic) Studies in Memoriam W. Vychichl, Leiden 2004.
BEESTON, A. F. L., A Descriptive Grammar of Epigraphic South Arabian, London 1962.
BEESTON, A.F.L., Sabaic Grammar, Manchester 1984.
COHEN, D., La phrase nominale et l’évolution du système verbal en sémitique; études de syntaxe historique, Leuven, 1984.
DILLMANN, A., Ethiopic Grammar, 2nd ed., London 1907.
EDZARD, D. O., Sumerian Grammar, Leiden 2003.
FAULKNER, R. O., The Ancient Egyptian Pyramid Texts, Oxford.
FLEISCH, H., Traité de philologie arabe, 2 vols., Beyrouth 1961-79.
GARDINER, A., Egyptian Grammar, 3rd ed revised., Oxford 1988.
GESENIUS, W. & E. KAUTZSCH, Hebrew Grammar, 2nd English edition by A.E. Cowley, Oxford 1910 (reprint: 1966).
GORDEN, C. H., Ugaritic Manual (UM), Rome 1955.
GORDON, C. H., Ugaritic Handbook, 3 Vols, Rome 1947
GORDON, C. H., Ugaritic Literature, Rome 1949.
HEINE, B. and NURSE, D. (eds), African Languages an Introduction, Cambridge 2000.
HETZRON, R. The Evidence for Perfect *y’aqtul and Jussive *yaqt’ul in Proto-Semitic, in Journal of Semitic Studies 14 (1969), p1-21.
HÖFNER, M., Altsüdarabische Grammatik, Leipzig 1943.
LAMBDIN, T.O., Introduction to Classical Ethiopic (Ge’ez), Harvard Semitic Studies 24, Ann Arbor 1978.
LIPIŃSKI, E., Outline of a Comparative Grammar of the Semitic Languages, 2nd Edition, Leuven 2001.
LOPRIENO, A., Ancient Egyptian : A Linguistic Introduction, Cambridge 1995.
LYONS, J., Introduction to Theoretical Linguistics, London 1971.
MOSCATI, S., An Introduction to the Comparative Grammar of the Semitic Languages, Wiesbaden 1964 (reprint: 1969 & 1980).
POPPER, K. R., Conjectures and Refutations, London 1972.
PRITCHARD, J. B., (ed.), Ancient Near Eastern Texts, 3rd ed. with Supplement, Princeton 1969.
RAINEY, A.F., Canaanite in the Amarna Tablets. A Linguistic Analysis of the Mixed Dialect used by Scribes from Canaan, 4 vols., Leiden 1996.
ROSENTHAL, F., A Grammar of Biblical Arabic, Wiesbaden 1961.
RUNDGREN, F., Das althebräische Verbum. Abriss der Aspektlehre, Uppsala 1961.
SEGERT, S. Altaramäische Grammatik, Leipzig 1975.
SIVAN, D., A Grammar of the Ugaritic Language, Leiden 1997.
SODEN, W. VON, Grundriss der akkadischen Grammatik, Roma 1952; Ergänzungsheft, Roma 1969.
THACKER, T.W., The Relationship of the Semitic and Egyptian Verbal Systems, Oxford 1954.
WATSON, J.C.E., The Structure of Mehri, Wiesbaden 2012.
WHITAKER, R. E., A Concordance of the Ugaritic Literature, Harvard 1972.
WOODWARD, R.J (Ed), The Ancient Languages of Syria – Palestine and Arabia, Cambridge 2008.
WRIGHT, W., A Grammar of the Arabic Language, 3rd ed., 2 vols., Cambridge (1962).
Bibliographical Abbreviations
AAG | SEGERT, S. Altaramäische Grammatik |
AE | LOPRIENO, A., Ancient Egyptian : A Linguistic Introduction |
ANET | PRITCHARD, J. B., (ed.), Ancient Near Eastern Texts |
ASAG | HÖFNER, M., Altsüdarabische Grammatik |
DGESA | BEESTON, A. F. L., A Descriptive Grammar of Epigraphic South Arabian |
EG | GARDINER, A., Egyptian Grammar |
GAG | SODEN, W. VON, Grundriss der akkadischen Grammatik |
GAL | WRIGHT, W., A Grammar of the Arabic Language |
GBA | ROSENTHAL, F., A Grammar of Biblical Arabic |
GKC | GESENIUS, W. & E. KAUTZSCH, Hebrew Grammar |
GUL | SIVAN, D., A Grammar of the Ugaritic Language |
ICE | LAMBDIN, T.O., Introduction to Classical Ethiopic |
OCG | LIPIŃSKI, E., Outline of a Comparative Grammar of the Semitic Languages |
RSEVS | THACKER, T.W., The Relationship of the Semitic and Egyptian Verbal Systems |
TSM | WATSON, J.C.E., The Structure of Mehri |
Footnotes
1 For a review of the various hypotheses see H. Fleisch, Traité de philologie arabe (1961-79), Vol.I, §126. See also E. Lipiński, Semitic Languages – Outline of a Comparative Grammar [OCG] (2001), §37/8, and S. Moscati et al, An Introduction to the Comparative Grammar of the Semitic Languages [ICGSL] (1964), §16.28 – 16.31. In general, Lipiński takes the Akkadian system to be primary whereas Moscati et al opt for an original based on the system in Classical Arabic.
2 For an overview of the category of aspect in general see J. Lyons, Introduction to Theoretical Linguistics (1971), p313 ff. For research in Semitic see Moscati et al, Introduction p131 ff ; Fleisch, Traité Vol II §123, and Lipiński, Outline, §38.3 ff.
3 An exception is F. Rundgren, Das althebräische Verbum. Abriss der Aspektlehre (1961), who bases his analysis on an initial distinction ‘stative vs fientive’ and then develops six sub-categories of the latter (§36). His theory is carefully worked out but seems implausibly complex.
4 See for example K. R. Popper, Conjectures and Refutations (1972), p33 ff and especially p53. See also On the Biradial Origins of the Semitic Triradical Root System (BOSTRS) §1.3. It will be claimed that there is an element of circularity in such an approach to diachronic linguistics. But this is to misunderstand the hypothetical-deductive method, which sets out to refute rather than ‘prove’ a hypothesis, with the consequence that a hypothesis is supported only to the extent that it is not refuted. However it is only fair to add that in the context of an essentially probablistic study, as this is, total refutation is impossible.
5 It is also possible to envisage an element <translative>, which would encode segments ‘cutting’ and ‘becoming clear’ of the event lines. This element is typically expressed in Semitic through active participle forms, marked for number and gender but not for person, and as such stand apart from full verb forms.
6 For example Lipiński, OCG, §38.4, who takes the primary distinction to be between <perfective> and <imperfective>, where the former is expressed by Akkadian iptaras and the latter by iparras. This is anyway a rather odd proposal as there is no comparable form expressing <perfective> e.g. in Arabic or the N. Ethiosemitic languages ; moreover he fails to show that apparent occurrences in the N.W. Semitic languages are primary, rather than secondary innovations perhaps under the influence of Akkadian itself.
7 The Arabic passive, and analogous forms in the other languages, are considered to be secondary innovations, albeit old, partly replacing, but to some extent replicating (or being replicated by), the various ‘Afroasiatic’ T- and N-forms (Lipiński, OCG §41.43 ff).
8 References to W. Wright, A Grammar of the Arabic Language [GAL] (1962).
9 For example, ‘if I see him (which I may) I will tell him’, as opposed to ‘impossible’, e.g. ‘had I seen him (which I did not) I would have told him’.
10 Egyptian also offers a degree of support for this conjecture. See A. Gardiner, Egyptian Grammar3 (EG) (1988), §150.
11 Wright, GAL, Vol II, §8 and §9. The ‘historical present’ usage cited by Wright is taken to be an innovation arising when temporal implications became superimposed on the original aspectual system.
12 Wright, GAL, Vol. I, §97. The shorter form appears to be less common ; in the Qur’ān the longer version occurs almost without exception.
13 Compare the dual forms, which end in –nni, e.g. (3p) yaqtulānni.
14 Wright, GAL, Vol II, §20.
15 See the discussion of Ugaritic verb forms at §2.3 above and in Section 5.
16 Cf. Lipiński, OCG, §39.5.
17 See W. Gesenius, W. & E. Kautzsch, Hebrew Grammar2 [GKC] (1966), §109.
18 Some investigators have argued that because declarative and non-declarative GPA forms display different stress patterns in the Masoretic text, stress in Common Semitic was phonemic; for example Lipiński, OCG, §25.8 and §38.2, and R. Hetzron, ‘The Evidence for Perfect *y’aqtul and Jussive *yaqt’ul in Proto-Semitic’, JSS 14 (1969), 1-21. This argument can be challenged on at least two grounds. Firstly, the reliability of the Masoretic pointing for the original pronunciation needs to be demonstrated rather than assumed (OCG, §7.5). Second, there are few if any contexts where declarative and non-declarative functions of the GPA form could be confused and therefore it needs to be shown that the differing stress patterns are primary rather than secondary. However, the phenomenon of distinct stress patterns is admittedly widespread and may be ancient.
19 See GKC §47m ff.
20 GKC §60e. Variant pronominal forms such as -ennu (3ms) are taken to be secondary innovations, although they also occur in Ugaritic and Aramaic (§2.3 and §2.4).
21 GKC §47m, §58i to l. and §61e
22 See A.F. Rainey, Canaanite in the Amarna Tablets: A Linguistic Analysis of the Mixed Dialect used by Scribes from Canaan (1996), Vol. II, p234ff. Rainey suggests apropos the el-Amarna texts (p236) ‘that the indicative energic in – una must be the ancestor of the Hebrew ... imperfect with accusative suffixes -enhu, ennu, etc’.
23 Canaanite, p242. This form is interesting in two respects. Firstly, Arabic provides very little evidence for energic derived forms. Second, derived forms in Akkadian proper distinguish <singulative> and <non-singulative> by apophony rather than through an agglutinated morpheme (-na) as appears to be the case here.
24 Lipiński (OCG, §5.1 ff) classes Ugaritic as a ‘North Semitic’ language but the evidence for so doing, rather than including it with the Canaanite languages, seems rather thin (§5.5). The second radical in Ugaritic verb forms on II-weak roots appears never to be written.
25 The myths and legends in these texts were intended for oral recitation and the texts thus exhibit stylistic and linguistic devices typical of such a medium - repetition, the use of rather formulaic phrases, and a general tendency to exploit the resources of the language for dramatic effect. For texts see C.H. Gordon, Ugaritic Handbook (1947), Vol. II, p159 ff. For translations see C.H. Gordon, Ugaritic Literature, (1949) and H. L. Ginsberg, in J.B. Pritchard (ed), Ancient Near Eastern Texts [ANET] (1969), p129-155.
26 The poetic texts are claimed to be significantly older than the prose texts and in consequence to have been less perfectly preserved; see for example D. Pardee, ‘Ugaritic’, in R.J. Woodward (ed), The Ancient Languages of Syria – Palestine and Arabia, (2008), p20. That the poetical texts are considerably older than the tablets on which they are preserved is likely, but for the purposes of analysis textual corruption must be regarded as an explanation of last resort.
27 Compare D. Sivan, A Grammar of the Ugaritic Language (GUL), 99.
28 One clear conditional construction is hm t’pn ‘l qbr bny tškṭ’nn bšnt ‘if they (vultures) fly over my son’s grave, rousing him from (his) sleep’ (1AQHT-3.151). The likliest interpretation of these verbs, assuming they are to be read as masculine forms (subject nšrm), is that they are 3mp GPE, perhaps vocalised *ta’apūn and *tušakṭa’ūnanu.
29 References are generally to the texts in volume II of Gordon, Ugaritic Handbook.
30 R.E. Whitaker, A Concordance of the Ugaritic Literature (1972), p492.
31 Whitaker (Concordance, p258) lists the root under ḥyy, but many of his citations support ḥwy, as for example àp ànk àḥwy àqht (2AQHT-6.32), where àḥwy appears to be a D-form.
32 Statistically, approximately 75 per cent of the examples on which this discussion is based support without difficulty the hypothesis ‘GPA ⇔ <singulative>’.
33 Ginsberg and Gordon (ANET and Ugaritic Literature) here utilise the present tense, presumably on the ground that the u in tbù must follow its aleph and therefore that the verb cannot be GPA.
34 Ginsberg and Gordon take both this expression and the similar qdš wamrr to represent single individuals and thus translate the verb as 3ms GPE. Compare àkr tmġyn mlàk ym ‘the messengers of Sea then arrive’ (137 30) where the verb is GPE.
35 Ginsberg and Gordon almost invariably translate such forms by the English present tense.
36 See for example C.H. Gordon, Ugaritic Manual (1955), §9.7.
37 The signs for ì and ù would appear to be later additions to the alphabet, in which case the sign for à would originally have been vowel free ; it may be that in some instances this remains the case (Sivan, GUL, 9, 16-18).
38 The Akkadian 3ms GPA form is imla. The infinitives of such verbs are of the form malū, analogous to those of III-weak verbs.
39 In addition to the examples cited, this discussion also draws on the following forms: ymġyn (1AQHT-4.170), ybln (2AQHT-5.12), tmtkṣn (‘NT-2.23), tḥspn (‘NT-2.38), àrgmn (‘NT-3.75), yqlṣn and ywptn (51-3.12), ttbn (51-7.24) and tmtn (125-3). Note in respect of ymġyn that the word order S-O-V seems to attract final –n even where the event is clearly <singulative>; compare ‘nt lbth tmġyn (‘NT-2.17).
40 Sivan, GUL, p102; Gordon, Manual, §9.8; Rainey, Amarna, Vol. II, p234 ff.
41 Gordon, Manual, §9.7; Sivan, GUL, p105 The form ìqrà is read *iqr’a by Lipiński (OCG, §39.10), but where, in any Semitic language, does a three-consonant cluster occur in a GP form?
42 Also t‘dbnh (3AQHT-4.33), yšlḥmnh (‘NT-1.5), yšqynh (‘NT-1.8), ìštynh (51-3.16), ykllnh and y‘msnh (51-5.72), t‘šlynh, tbkynh and tqbrnh (62-1.15).
43 Compare also tškṭann [3mp with suffix –n] (1AQHT-3.151), tbrknn (2AQHT-1.24), t‘nynn (2AQHT-6.32), tštn (3AQHT-4.27 and 128-24), tbq‘nn etc. (49-2.26), tštnn (62-1.15), ywsrnn (127-25). Form tlùàn (KRT-33) displays suffix –n but is otherwise difficult to analyse (GUL, p103). Note that –nh never occurs with root šyt ‘place’, suggesting that some of these associations are merely conventional.
44 Also the following plural forms: t‘pn (1AQHT-3.151), ytnn (2AQHT-5.26), y‘nyn (‘NT-4.49), tqṭṭn utkṭin (2-1.15), t‘rbn (5-9), y‘tqn (49-2.26), yštn (51-4.14), tnġṣn (68-17), tikln and tqrṣn (75-1.10), tmtn (125-105), tbùn (128-4.21), tbkn (128-5.12). Reasonably clear dual forms are: tmġyn and tṣḥn (1AQHT-2.89), tmġyn (49-1.31), tmgnn and tġẓyn (51- 3.25), tṣḥn (52-42), tqtnṣn and tldn (52-51).
45 GUL, p98.
46 This position contra Lipiński, OCG, §25.8 and §38.2.
47 Sivan (GUL, p102) concedes that ‘it is exceedingly difficult to identify such forms in Ugaritic’.
48 ‘Since most of the examples in Ugaritic are not from verbs with third aleph, one cannot know if they are the energic of the injunctive or of the indicative’ (GUL, p105).
49 GUL, p105; Ugaritic Language, p59. In Ugaritic Manual, §9.7, Gordon translates these forms in the same way as Sivan. Sivan also refers to a form yṣàn which is not in Whitaker’s concordance. Gordon contrasts forms yspì (121 2.10) and ìspà (49 5.20) on root spà ‘eat’. Unfortunately the adjacent text is in both cases damaged. The ì of yspì is the more difficult for if, as seems likely, the theme vowel of a III-aleph root was a, the ì becomes inexplicable.
50 The vocalisation *yimġū might be expected for a GPA form.
51 A more convincing example is the form ùbà on root bwà ‘enter’ and contrasted with ùbù, cited in D. Pardee, ‘Ugaritic’ in R.J. Woodward (ed), The Ancient Languages of Syria – Palestine and Arabia (2008), p24. Pardee and Sivan (GUL, p105) vocalise ùbà as *’ubū’a.
52 The term ‘Old Aramaic’ is used as in S. Segert, Altaramäische Grammatik [AAG], (1975).
53 It may be that other examples have been identified in the years since Segert published his work.
54 Segert, AAG, §5.7.6.4.5, p288 and §5.7.6.9, p291.
55 In this respect Aramaic resembles Arabic and Hebrew, although the n appears more consistently in Aramaic than in Hebrew. For Biblical Aramaic paradigms see F. Rosenthal, A Grammar of Biblical Aramaic [GBA] (1961), p60.
56 See the summary at AAG, §5.7.1.1.9 (p266). The final n is always omitted in the Ya’udi dialect (AAG §5.6.4.7.2) so that in this dialect it is not possible to tell by inspection whether the relevant forms are GPA or GPE.
57 Rosenthal, GBA, §175 and p71.
58 But see the forms displaying suffixes without –n at AAG, §5.7.9.4.4 (p310). Most of these are cited without context so that, although ’ḥṣlk ‘I will rescue you’ appears to be functionally <non-singulative> others may be GPA.
59 Recall that Hebrew GPA (jussive) forms can occur either in the protasis or apodosis, or both, of ‘possible’ conditional clauses (GKC, §109 h/I and §159).
60 For paradigms see W. von Soden, Grundriss der akkadischen Grammatik [GAG] (1952; Supplement 1969), Verbalparadigmen, 9*. Excepting the 3fp and 2p forms this paradigm is almost identical to that of the Arabic majzūm (Wright, GAL, Vol. I, p298).
61 Unlike Arabic and Hebrew, where GPA forms appear both in protasis and apodosis of ‘possible’ conditional constructions, in Akkadian they are confined to the protasis. Also in contrast to Hebrew and Arabic, Akkadian GPA forms occur in the protasis of ‘impossible’ conditions (GAG, §161c, §162a).
62 GAG §83 and §82 respectively.
63 GAG, Verbalparadigmen 9* note 2.
64 GAG, Verbalparadigmen 8* and 10*
65 GAG, §78.
66 A. Dillmann, Ethiopic Grammar (1907), §90.
67 This presumed closeness between Ge‘ez and Arabic is not supported by the differing 1st and 2nd person subject pronouns in the Arabic and Ge‘ez suffixing verb forms. Also the Ge’ez GPA form does not occur in conditional clauses (T.O. Lambdin, Introduction to Classical Ethiopic (ICE) (1978) §51.1, p231).
68 Root ’ty, cited from M. Höfner, Altsüdarabische Grammatik (ASAG) (1943), §64. See also N. Nebes and P. Stein, ‘Ancient South Arabian’ in R.D. Woodward (ed.), The Ancient Languages of Syria-Palestine and Arabia (Cambridge 2008), p157, who designate this example as ‘Early Sabaic’.
69 See A.F.L. Beeston, Sabaic Grammar (1984), §5.12.
70 ‘South Arabian’, p155. See also Beeston, Sabaic Grammar, §5.7.
71 Höfner, ASAG, §59 table 7; Nebes and Stein, ‘South Arabian’, p163. Höfner takes the Arabic system to be primary and attempts to analyse the ESA GPE forms along the ‘traditional’ dimension ‘imperfect’ (her Table 6) vs ‘energic’ (Table 7). Although in respect of the patterning of its consonants the Minaean 3fp form in Table 7 certainly matches the Arabic 3fp energic form taqtulnānni, Höfner has to recognise (§60) that, as in Ugaritic, many forms in final n do not have ‘energic’ implication.
72 Höfner (ASAG, §59) associates the second –n with that occurring in the infinitive of derived stems. See also Nebes and Stein, ‘South Arabian’, p158 and A.F.L. Beeston, A Descriptive Grammar of Epigraphic South Arabian (DGESA) (1962), §19.
73 ‘South Arabian’, p156. Their description uses f‘l as the base root.
74 Cited from §4.4.2.2 in ‘South Arabian’ (p156)
75 Nebes and Stein ‘South Arabian’, p157. But note that Beeston (DGESA, §21-9) draws attention to Sabaic GPE forms used in jussive constructions.
76 ‘South Arabian’, p168; DGESA, §21-9.
77 The ‘v’ in GVP symbolises a long vowel within the root pattern.
78 Data from Johnstone, Mehri Lexicon pages xxi/xxii. The passive subjunctive and imperfect are not absolutely identical but clearly share a common origin.
79 See §2.6.1 above and J.C.E. Watson, The Structure of Mehri (TSM) (Wiesbaden 2012) §2.5.1.3.2.2.
80 Recall that verb forms in final -n can occur in the apodosis of ‘possible’ conditional clauses in ESA (Nebes and Stein ‘South Arabian’ p168).
81 Similar forms also occur in Berber (See Berber ; a Semitic Language? (BeSL) §1.1) and Egyptian (§6.2.1 above).
82 Arabic and Biblical Hebrew have a number of D-form verbs without an accompanying G form (Arabic approximately 4.0%, Hebrew 4.5%). Some of these may earlier have been accompanied by, or perhaps have originated in, a G form, but it is not impossible that others originate in some form analogous to iparras. See Lipiński, OCG, §38.7, Fleisch, Traité, Vol. II, §126p and Cohen, ESVS p65/9.
83 Cohen (ESVS p67) also discusses the possibility that forms such as yənaggər result from the re-assignment of a (jussive) D form.
84 The presence of theme vowels in GPG forms is predicted by the rules propsoed for the evolution of these forms in §8.6 of Towards a Morphology of the pre-Semitic Verbal System (MPSVS). Nonethless it cannot be shown unambiguously that these vowels are primary rather than secondary, in the latter case having evolved by analogy with the associated GPA forms.
85 Biblical Hebrew also has very few (about 8, approximately 0.7% of the total). Arabic has rather more (about 2.8% of the total) and Egyptian about 5%. They appear to be considerably more common in Ge‛ez and perhaps also in Mehri (no statistical analysis carried out for either language). Note that the ‘imperfect’ of Ge‛ez reduplicating verbs is of the form yəbadabbed
86 For Sumerian GPR forms see D.O. Edzard, Sumerian Grammar, §12.4.3 (2003) ; is it conceivable that Sumerian reduplicating forms were the source of the Akkadian GPG form? The conjecture that a GPR form expressing <non-singulative> aspect was a feature of Common Semitic entails that, where preserved, the forms and functions of such verbs at some time evolved to yield a suffixing form to express the associated <singulative> aspect.
87 See for example GKC, §48 and §58 for Hebrew, and for Ugaritic, Sivan, GUL p102.
88 Rainey, Canaanite, Vol. II p235. Like Sivan for Ugaritic (Table 1 above), Rainey attempts to distinguish between ‘indicative’ and ‘injunctive’ energics but can adduce almost no evidence in support of the latter (p263). As with Sivan, Rainey’s overall scheme reflects more a desire for symmetry than respect for the evidence of the texts.
89 For Arabic, this conclusion would of course imply that the final morpheme -n(a) of the relevant mudāri‘ forms and the energic morphemes have similar origins.
90 Höfner, ASAG, p74. See also Nebes and Stein, ‘South Arabian’, §5.3.6, §5.4.1 (30B), §5.5, §5.6.1.
91 Nebes and Stein, ‘South Arabian’ §5.3.6.
92 Note that this form more closely resembles the shorter Arabic energic form with suffixed -a/un than the longer form in -inna/unna.
93 For the variant Arabic energic paradigms see Wright, GAL, Vol I p298. For el-Amarna see Rainey, Canaanite, Vol. II p228/34.
94 See the table at Lipiński, OCG, p379.
95 Canaanite, Vol. II p262.
96 GUL p104.
97 See the introductory remarks in Lipiński, OCG, p24. Certain of the differences between Egyptian and Semitic become more explicable if a degree of relationship is postulated between Egyptian and the relevant African languages, especially Cushitic.
98 This section draws primarily on Gardiner’s Egyptian Grammar [EG] and T.W. Thacker, The Relationship of the Semitic and Egyptian Verbal Systems [RSEVS] (1954). Compare the syntactically-based treatment in A. Loprieno, Ancient Egyptian : A Linguistic Introduction [AE] (1995), 4.6.1 to 4.6.3.
99 Gardiner, EG, §20. Loprieno takes the value of the character usually transcribed in English by ì always to be /j/ (= /y/).
100 Compare for example, Egyptian ’m ‘swallow’ and km ‘not know’ with Beḍawiē ’ām ‘eat’ and gam ‘not know’ (assuming these are not Egyptian loans into Beḍawiē). Among nouns, Egyptian sn ‘brother’, for instance, has no Asiatic equivalent but is identical to Beḍawiē san and to cognate forms in other Cushitic languages and Chadic. Thacker is of the opinion that apparent biconsonantals are in fact II-weak triradicals with abbreviated orthography (RSEVS, p52ff). A sample of 848 roots, drawn from R.O. Faulkner, A Concise Dictionary of Middle Egyptian (Oxford 1962) and Gardiner’s Egyptian Grammar, yielded 112 biconsonantal (14 per cent of the total) and 25 geminate roots (3 per cent). In contrast, geminate roots in Arabic comprise about 7 per cent of the total.
101 See for example section §2.5 in D.L. Appleyard ‘Beja as a Cushitic Language’.
102 For Egyptian paradigms see EG, §39 and §67.
103 Mutable verbs are those where the stem phonemes are written differently, according to context (EG, §268). Geminate roots, for example, can be written with one geminate consonant or two, as in qb.n.f or qbb.n.f, the latter spelling perhaps reflecting doubling of the first geminate consonant (§269). See also AE, p79, EG, §410/13. The śdm.n.f ‘imperfective’ form is taken to signal repetition or continuity (EG, §295).
104 See for example Lipiński, OCG, p25 ‘... it stands to reason that Egyptian ... lost the prefix-conjugation in prehistoric times’.
105 RSEVS, p224. Compare section §4.1 in MPSVS.
106 RSEVS, p234.
107 EG, §411. Gardiner conjectures that the śdm.f and śdm.n.f forms evolved in parallel with or by analogy with the ‘relative’ form (EG, §380-389). His conjecture is carefully argued but not entirely convincing. For the final w characteristic of the passive participle, and occurring at least occasionally in the ‘relative’ form, appears to be entirely absent from the śdm.f and śdm.n.f forms, with the occasional exception in the passive śdm.f (EG, §420). It could also be that relative forms evolved from passive participles by analogy with śdm.f and śdm.n.f, rather than the reverse.
108 EG, §411.1.
109 EG, §64; Moscati et al, Introduction, §13.1.
110 EG, §43. An alternative hypothesis could be that genitival suffixes were added to the independent pronouns by analogy with the verb subject pronouns.
111 For the full functions of the śdm.n.f form see EG, §411.2 and §414. For the śdm.f form see EG, §450.
112 EG, §450.1. Aside from negatives, Gardiner gives no example of an śdm.n.f form expressing other than past time. Certain forms in the Pyramid Texts could be interpreted as expressing future time, as for example rd.n ḥrw in lines 24c/d, which Faulkner translates as ‘Horus has caused’ but whose context would appear equally to permit ‘Horus will cause’. See also Loprieno, AE p77.
113 EG, §455 and §418 respectively.
114 Note also the verb wn ‘be, exist’, where wn (śdm.f) expresses past time or non-duration and wnn (śdm.n.f) expresses future time or duration (EG, §107.1, §157).
115 EG, §449.
116 However, this characterisation of the function of the śdm.f form cannot easily be reconciled with that postulated by Gardiner for the śdm.n.f form, namely ‘...to present the verbal action as an incident, as something happening or occurring to someone, irrespective of time position.’ (EG, §411.1).
117 EG, §150/1. Compare the use of Arabic energic forms in ‘future’ conditional constructions.
118 See for example Moscati et al, Introduction, §16.31. However, for Egyptian, Gardiner states that ‘...in an early lost stage of the Egyptian language, the old perfective was a freely used narrative tense with both active and passive meanings’ (EG, §311) but then notes that this usage ‘survives only in a few patently archaistic examples’ and that generally ‘the old perfective from….transitive verbs has passive meaning’. For paradigms see EG, §309 and GAG p8*. Compare also ‘qualitative’ verbs in Berber (BeSL §2.5)
119 But note that conditional constructions similar to those of Akkadian, Hebrew and Arabic also occur in Berber (BeSL §9.2) and in the Cushitic-Semitic language Beḍawiē (see Beḍawiē as a Semitic Language §3.2).
|