Evidence of an early metathesis among Akkadian piristum-stem nouns.

• Author: Testen, David

1. INTRODUCTION

   Discussions of Semitic grammatical issues frequently fail to draw a distinction between two discrete aspects of the investigation of morphology, viz. synchronic parsing and diachronic etymology. (1) The core of the typical Semitic word-stem is said to lie in the root, an ordered set of consonantal elements serving as the word's formal and semantic skeleton. The root serves as the "raw material" out of which the morphological machinery of the language "constructs" the stem in question. Identifying the root within a given stem is, in one sense, a fairly simple question of parsing. There is, however, an intrinsic duality to the role of the root within the methodology of grammatical analysis--while, on the one hand, the root serves as the origin of the word-stem, at the same time it is the product of the analyst's act of grammatical deconstruction. As long as we are dealing solely with the synchronic plane, the difference between these two aspects of the root--the root as a building block in the construction of words vs. the root as an artifact of grammatical analysis--is fairly trivial. The distinction becomes crucial, however, when the analysis shifts to the level of diachrony, where our principal concern lies in the explication of linguistic history. Once we allow for the possibility that language change might play a key role in determining the shape which any given word has assumed by the time of its documentation, it becomes apparent that it is impossible to draw a simple equation between that word's synchronic root--the set of consonants which the grammatical parsing of a given stem will extract--and its etymological root. Since linguistic formations are inherited from the past at least as much as they are generated in situ in the present, an etymological analysis will often demonstrate that, in historical terms, a given linguistic formation is ultimately the product of a set of developmental factors quite different from the grammatical processes of the language in which it is embedded.

2. A HYPOTHESIS

   The following pages are intended as a case study, drawn from the prehistory of Akkadian, of the significance of the distinction between the root accessible to a "parsing" and the root unearthed through an etymological analysis. The material under consideration here is the set of Akkadian nouns of the shape ([C.sub.1])i[r.sub.2]i[C.sub.3]tum, i.e., feminine nouns of the pattern "piristum" in which the second radical is -r- and the third radical is non-weak. An examination of the recent Concise Dictionary of Akkadian (2) provides us with twenty-eight instances of words of the shape (C)iriCtu(m) and (C)eriCtu(m). Of these, one at least ("miriqtu II") is probably best discarded, (3) and we should allow for the possibility that several of the remaining words actually reflect other formations--at least one of them (erimtu II "cover(ing)") is taken by von Soden as a reflection of the stem-shape *CariC-t-um, (4) two others (eristu | "cultivation" and giristu "loaf of bread") seem to contain an underlying long vowel, (5) and two further stems (pirindu and tirimtu) are liable to have yet other origins. (6) The remaining twenty-two words are listed in Table 1, along with any other terms with which the Akkadisches Handworterbuch (AHw) or the Chicago Assyrian Dictionary (CAD) associates them. Since there can be little doubt that the ongoing investigation of the Akkadian language will continue to unearth lexical items of this structure, we cannot treat this list as comprehensive. (7) Nevertheless, this list will serve as an adequate basis for the present investigation.

   The claim will be made here that, in etymological terms, the members of this set of nouns may be traced back to two distinct root-types. While, for the majority of these nouns, the historical root proves to be the same as the root revealed by a synchronic parsing (viz. *[square root of ([C.sub.1][r.sub.2][C.sub.3])]), it is suggested that in a small but significant number of cases we find words which are better traced back to etymological roots in which the -r- was the third radical rather than the second (*[square root of ([C.sub.1][C.sub.2][r.sub.3])]), and that the forms in question seem to result from a metathesis which, at an early stage in the prehistory of Akkadian, reversed the order of the last two radicals. It is important to note that the hypothesized metathesis is envisioned as having operated specifically within the nominal stem-formation piristum (i.e., [C.sub.1]i[C.sub.2]i[C.sub.3]-t-um)--or, to be more precise, in the pre-Akkadian forerunner of this stem-type--rather than as operating at the more nebulous level of the abstract root. It is also worth noting that the potential relevance of the posited metathesis ultimately extends well beyond these specific forms.

3. THE SYLLABIC CONFIGURATION OF EARLY SEMITIC

   The reference works on Akkadian grammar routinely describe the second -i- of the stemtype piristum as a secondary insertion, an epenthesis rendered necessary by the complexity of the cluster which would otherwise result from the addition of the feminine suffix -t- to a stem ending in a cluster (pirs-). The recent Structural Grammar of Babylonian of Buccellati (8) phrases this development in the following terms, and substantially similar analyses may be found in earlier studies: (9)

   ... Following morphophonemic rules ... when -t is added to the patterns pirs- and purs-, an auxiliary vowel -i is inserted between the second and third consonants of the pattern, e.g., {sikl-t-um} /sikiltum/ "gain" or {rutb-t-um} /rutibtum/ "swamp" ... (10) While these analyses are framed in synchronic, structural terms (i.e., they are intended as characterizations of the derivational processes by which Akkadian generates a form of the shape piristum out of the underlying constituent elements *pirs- and *-t-), the notion that the shape piristum owes its second vowel to a secondary process seems to be equally valid on the diachronic, historical plane. There is a good likelihood that, at a prehistorical stage in the development of the Akkadian language, the ancestor of a stem of the piristum-type had no vowel in this position (*pirst-um or *[C.sub.1]i[C.sub.2][C.sub.3]tum), and that, in the course of the evolution of the early "pre-Akkadian" language into the documented, historical language, an epenthesis took place which had the effect of disrupting the original triconsonantal cluster through the insertion of an epenthetic vowel -i- (*pirst- > pirist-). This is made clear through a comparison of the corresponding structures evidenced by the related Semitic languages. Among the West Semitic languages we see no sign of a stem-shape *[C.sub.1]i[C.sub.2]i[C.sub.3]-t- as the counterpart to Akkadian pirist-: in Arabic we find rather the stem-shape [C.sub.1]i[C.sub.2][C.sub.3]at- (e.g., fikrat-un "idea," tiflat-un "young one (fem.)"), in Hebrew [C.sub.1]i[C.sub.2][C.sub.3]a (e.g., simha "joy," '[epsilon]gla "heifer"), and in [TEXT NOT REPRODUCIBLE IN ASCII] ([TEXT NOT REPRODUCIBLE IN ASCII] "drunkenness," [TEXT NOT REPRODUCIBLE IN ASCII] "filling up (n.)"). Such forms suggest that feminine-suffixed counterparts to the stem-shape *[C.sub.1]i[C.sub.2][C.sub.3]- in early West Semitic employed the vowel-bearing suffix *-at- rather than *-t-, and that consequently no epenthetic vowel was inserted between the second and third radicals. (11) Since the two principal branches of Semitic have adopted quite different strategies in shaping the feminine-suffixed counterpart to *[C.sub.1]i[C.sub.2][C.sub.3]-, there is every reason to suspect that Akkadian piristum represents an East Semitic innovation and thus that, at the time of the ancestral language, syllabic configurations of the shape *[C.sub.1]i[C.sub.2][C.sub.3]-t- did not yet pose a phonotactic problem (fig. 1).

   To be sure, the reconstructed configuration *[C.sub.1]i[C.sub.2][C.sub.3]-t-um runs counter to certain preconceptions about the nature of Semitic syllables. Our oldest evidence for the Semitic languages comes from languages which display relatively impoverished sets of allowable syllable structures--the phonotactic system of historical Akkadian, for example, permits only single-consonant syllable onsets (i.e., syllables of the shape CV- but not, e.g., *CCV-) and only syllables ending in a short vowel (CV-), a long vowel (CVV-), or a short vowel followed by a single consonant (CVC-). Whether or not the Semitic ancestral language showed the same restrictions in its syllable system, however, is a question quite distinct from whether or not the descendant languages do. It is clear that, like any other linguistic feature, the set of allowable syllable structures for any given language is subject to change over the course of time--to take an example from Semitic, compare the wide range of syllable-types employed by the modern Maghribi Arabic dialects in contrast to the much more limited array of syllable-shapes displayed by standard Arabic (cf. Moroccan [TEXT NOT REPRODUCIBLE IN ASCII] "(I) wrote" vs. standard Arabic katabtu).

   [FIGURE 1 OMITTED]

   Comparative Semitic data provide us with reason to suspect that the picture that has been fashioned for the Proto-Semitic syllable, based on the model provided by the languages of classical philology, is unduly simplistic. The juxtaposition of a set of forms such as the familiar Semitic terms for "name" (e.g., Akkadian sum-um, Arabic ('i)sm-un, Hebrew sem) is instructive in that, while each of these shapes conforms to the prevailing notions of allowable Semitic syllable configurations, each does so in a different manner--the Akkadian form avoids an initial cluster through the presence of an -u-, the Hebrew form by means of the reflex of an earlier *-i-, and the Arabic form by means of a preposed syllable ('i-). In the absence of a principled analysis capable of demonstrating that any one element of this set (Akkadian -u-, pre-Hebrew *-i-, or Arabic...