The Morphology of the G-Stem Imperative in Semitic.

• Author: Bjoru, Oyvind

1. INTRODUCTION

   The morphological nature of the G-stem imperative in Semitic remains a contentious problem to which multiple incompatible solutions have been proposed. In this paper, I will assess earlier approaches and proceed to reconstruct three disyllabic imperative patterns for Proto-Semitic, i.e., qutul, qitil, and qital.

   We can trace three explanatory strains in the literature on the morphology of the Semitic imperative. First, it has been claimed that the imperative is synchronically clipped from the verb in the prefix conjugation, and that the resulting form is then adjusted to fit the syllabic structure of each language. Alternatively, the clipping is situated at a specific node in the Semitic Stammbaum or as a process within Proto-Semitic. Second, a verbal base that does not appear as a full word itself could be used to build both the prefix conjugation and the imperative. And third, one or several designated vocalic patterns could be employed specifically for the imperative. I will begin by refuting the former two propositions before promoting a version of the latter I find to be the most economical and precise in accounting for the realizations of the imperative in the various branches of Semitic. In doing so, I try to account for as many forms as possible while avoiding the introduction of ad hoc sound changes.

2. THE PREFIX CONJUGATION HYPOTHESIS

   Most grammars of individual Semitic languages derive the imperative from the prefix conjugation by shedding the prefix and adjusting the resulting base to conform to the permissible syllabic structures in the language, e.g., Akkadian preterite i-prus > **prus > imperative purus. (1) In West Semitic, the older and shorter form of the prefix conjugation, preserved as a jussive, is taken to be the starting point of the process, e.g., Arabic jussive yaktub > **ktub > imperative (u)ktub. Languages would differ as to how they resolve the CCVC that results from dropping the prefix when the root contains no weak consonants, either inserting a vowel to expand the cluster into a syllable (Akkadian), or prefixing a vowel when necessary (Classical Arabic). Whether such a derivation is a mere didactic tool or taken to be a synchronic process in either the language in question or one of its ancestors is typically not stated explicitly in textbooks or teaching grammars. (2) Addressing the morphology synchronically or with a view to diachrony are fundamentally different approaches. A synchronic analysis is no less valid than a diachronic comparison and reconstruction, but we must engage in the latter if we are to tackle questions of how the imperative in Proto-Semitic would have looked, and what routes of development are most plausible for the daughter branches and languages.

   2. Kienast held that clipping the prefix conjugation is the common Semitic formation of the imperative, and he took the various ways of resolving the resulting initial consonant cluster in different languages as evidence that this morphological derivation is both original to Proto-Semitic and ongoing in the daughter languages. (3) As the argument goes, the Arabic prosthetic vowel (u)qtul is a fundamentally different form from the Akkadian purus, and they can therefore only be synchronic solutions to an impossible CCVC arising when the prefix is clipped and not the result of regular sound changes affecting an original CVCVC pattern.

   Admittedly, clipping the prefix conjugation would produce most imperative forms in any Semitic language, but in certain cases the imperative, by virtue of its different syllabic structure from that of the prefix conjugation, retains phonological material lost in the prefixed forms. The clipping would then have to be located at a specific proto-stage of the language as it develops. Hebrew I-? verbs are a case in point. The prefix conjugation yo(?)kal does not correspond to an imperative **kal, but to [phrase omitted]. Such a form would have had to be clipped before the Proto-Northwest Semitic (Proto-NWS) form *ya?kul underwent the following string of changes: the quiescence of the aleph (*ya(?)kul) and lengthening of the prefix vowel (*yakul), the Canaanite shift a > 5 (*yokul), the dissimilation of the second of two rounded vowels (d/u) to i (*yokil), and the shift of the rare i-class verbs to a (yokal). And even then, the initial cluster in **Pkul must be broken up by a hatef-vowel. This effectively puts the terminus ante quern for the clipping in Proto-NWS. And in that case, the cognate Arabic kui further muddies the water, demanding that we posit either a NWS innovation reinstating root integrity or a separate Arabic innovation, shedding the initial glottal stop. In any case, for this explanation to work, the clipping of Pekol would have to be a process operating in NWS, which then comes to an end, freezing the form that is in turn inherited by Hebrew. (4)

   Similarly, the Harari reflex of [??]bhl 'to speak' is yal (< *yabal) in the jussive, but bal in the imperative, preserving the initial etymological root consonant rather than a clipped **al. If the form was ever clipped from the prefix conjugation, the process must have stopped before intervocalic b was lost in this verb. (5)

   With Akkadian I-n verbs, we even get forms that cannot be reconciled with clipped prefix conjugations at an earlier stage of the language. Synchronically, preterite iddin should presumably produce the imperative din 'give!', which happens to be the regular form in Assyrian. The Babylonian imperative idin, on the other hand, cannot come from a verb in the prefix conjugation. Proponents of the clipped imperative would say that copying the theme vowel onto the initial syllable is the way Akkadian resolves the impossible forms CCVC synchronically, (6) but with din, the process is not necessary, and if anything, we might expect **didin from **-ddin, and not idin.

   It is my contention that simply positing a process that clips a CCVC base from the prefix conjugation and resolves the cluster is wholly insufficient to explain the imperative forms we encounter. It cannot obviate the need for a conscientious reconstruction.

   The only way to salvage such an explanation of the morphology would be to pinpoint at which stages of the various branches of Semitic the process stopped, or to claim that clipping and cluster resolution is strictly Proto-Semitic, resulting in fixed patterns separate from the prefix conjugation that then undergo regular sound changes, producing the forms we are presented with. The latter scenario is compatible with what I am arguing for here, i.e., qutul, qitil, and qital as distinct imperative patterns from Proto-Semitic onward, but it seems arbitrary to insist on the additional step of deriving these from the prefix conjugation as long as the process would be internal to Proto-Semitic and no additional evidence is introduced.

   Besides the morphological problems with deriving the imperative from the prefix conjugation, the semantic motivation is less than obvious. From a West Semitic perspective, a relationship between the jussive and the imperative is perhaps not wholly unimaginable, as both are modal forms, but in East Semitic, the short prefix conjugation is an indicative preterite. N. J. C. Kouwenberg claims that the secondary irrealis function of the iprus connects it with the imperative. (7) The iprus is admittedly used to form the precative form liprus, but it is first and foremost the prefix la- that marks the form as injunctive. (8) Hence there does not seem to be a strong modal association between the preterite and imperative, and it cannot buttress the view that the forms must be connected. At best, the semantics are not necessarily an obvious impediment to such a derivation.

   While there is reason to dismiss clipping as a general rule to derive the imperative from the prefix conjugation in the Semitic languages, we may suspect that analogical realignment of the imperative forms to the prefix conjugation has occurred on occasion in certain verb classes. (9) This ought, however, to be an explanation of last resort and not a procrustean bed on which all imperatives must be made to conform.

3. THE VERBAL BASE HYPOTHESIS

   Lipinski holds that **qtVl served as a basic building block for the prefix conjugation while being used as the underlying form for the imperative. (10) That is to say, a bound morpheme **qtVl was combined with personal markers to form the prefix conjugation and underwent expansion to a disyllabic form by copying its only vowel to build the imperative, i.e., qtVI > qVtVl. It is unclear what reality this model ascribes to such bound morphemes. Are they lexically stored as such? To my mind, the **qtVl rather resembles an abstraction that springs from analysis.

   Lipihski's pattern, as it stands, might also violate the standard syllabic structure of Proto-Semitic, which seems to avoid initial consonant clusters. The patterns CV, CV, and CVC are the only licensed syllables in the classical Semitic languages, and the same set must be reconstructed for Proto-Semitic. (11) D. Testen has, however, tried to demonstrate that the reflexes of certain basic lexemes are most efficiently accounted for by positing syllabic sonorants for Proto-Semitic, e.g., *bn 'son' (Hebrew ben, Arabic ibn-), *[theta]n 'two' (Akkadian sin-, Arabic i[theta]n-), effectively allowing some such clusters in the proto-language. (12) He goes on to expand the argument to certain particles, e.g., precative *l- in Akkadian l-iprus and Arabic wa-l-yaf[??]al. (13) Furthermore, Testen explains the loss of initial n- in Akkadian I-n imperative forms as a resolution of forms like *n[theta]ur > usur 'guard!' and *nkis > ikis 'cut off!'. (14) It appears to be his assumption that CCVC, without resolving the initial cluster, is the imperative pattern that should be reconstructed for Proto-Semitic. Bar-Asher points out, however, that if the imperative begins with such a cluster, Il-n verbs in Aramaic...