No CrossRef data available.
Article contents
Reflexes of extinct phonemes in Semitic1
Published online by Cambridge University Press: 24 December 2009
Extract
Phonological distinctions vanish in the course of the evolution of a language, but in diachronic phonology it is known that the distinctive oppositions involved do not necessarily disappear as a consequence thereof. Certain effects brought about by the ancient phonemes while they were still articulatorily different do survive after the originally distinct articulations have merged into one. The surviving effects are customarily called ‘reflexes’, that is, of the phonemes now extinct, and it is they that permit us to assume with a considerable degree of certainty that the distinctions now perceivable by virtue of the surviving effects were once vested in different phonemes in the same language. The doctrine of phoneme reflexes has contributed immensely to an ampler, more complete, and more refined knowledge of the sound system of the ancient Indo-European languages. In Semitics, however, this line of investigation has hitherto been much less exploited. While there can be no doubt that we have a fairly good knowledge of what might be termed ‘the common Semitic sound system’, such insights can only tell us what phonological distinctions exist in one historical Semitic language, while being absent from another, and how correspondences should be drawn up between them. The advantage, however, of the study of reflexes is that it affords insights into the mechanism of one and the same language and permits inferences on historically unattested distinctions as well as the reconstruction of a stage of a language prior to its earliest written documentation. We can grasp here the real difference between comparison and reconstruction: while comparison enlightens us as to how languages differ, it cannot make us see the true identity or physiognomy of any given language. This latter task is dependent upon the doctrine of reflexes. If certain otherwise common Semitic distinctions involving pharyngals are reflected in Early Akkadian as neatly regulated mechanisms of vocalic patterns, we must infer that it is not true to say that Akkadian is set apart from other Semitic languages by the absence in it of pharyngal articulations, because these articulations, since they left traces, could not have been absent in that language in its earliest stages. The identity of Akkadian will, therefore, have to be established on the ground of the presence, rather than the absence, of a certain statable number of pharynsal phonemes.
- Type
- Articles
- Information
- Bulletin of the School of Oriental and African Studies , Volume 41 , Issue 3 , October 1978 , pp. 443 - 452
- Copyright
- Copyright © School of Oriental and African Studies 1978
References
2 Thus, if the differentia specifica between e and a be given phonemic status and symbolized, say, by /E/ (materialized suprasegmentally), a form such as ešēru ‘be straight’ can be synchronically represented as /Eašāru/ distinct from /'ašāru/ ašāru ‘locality’ as well as from /wašāru/ (w)ašāru ‘be released’; and Akk. /E/ would then be a phoneme inter alia corresponding to Heb. y. We can likewise posit Akk. /:/, e.g. in še-mu-u represented as /šamu:E/ ‘hear', where /:E/ corresponds to final’, ḥ of other Semitic languages.
3 /E/ as a successor to consonantal ḥ would be, in a transitory stage, a concomitant of that phoneme still represented in writing by (the approximation?) ḫ; thus, ‘to blow, kindle’ (with Semitic ḥ) appears as napāḫu as well as nepēḫu; that is, /ḫ/ corresponds to ḫ, and /E/Øḫ/ corresponds to ḥ. My colleague Aaron Shaffer also informs me that, if the Old Akkadian spelling il-qa-a were to be read (rather than as ilqâ, a variant of more common ilqê, see Chicago Assyr. dict., IX (L), 131a) as il-qa-aḫ, it would be written evidence corroborating the above inference, namely that /:E/ succeeded written ḫ as a representative of Semitic ḥ in final position.
4 Studies in Semitic grammar and metrics, Wroclaw, Polska Akademia Nauk, 1973, ch. i, ‘The verbal root in Semitic’Google Scholar (reviewed in BSOAS, XXXVII, 2, 1974, 449–50).Google Scholar
5 cf. Bergsträsser, G., Hebräische Grammatik, I, Leipzig, 1918, 88Google Scholar, for a view combining ‘Wurzelvariation’ with possible phonetic developments.
6 Kurylowicz, , op. cit., 19 f., 23 f.Google Scholar
7 cf. ‘An outline of a general theory of juncture’, in Rosén, H. B. (ed.), Studies … in honour of H. J. Polotsky, Jerusalem, 1964, 162, 171.Google Scholar
8 op. cit., 28 f.
9 cf. Brockelmann, , Grundriss, I, Berlin, 1908, 238 fGoogle Scholar. It may be not altogether useless to develop here some more arguments which tend to show why a dissimilatory explanation is, in cases of this nature, preferable to an assimilatory one. For example, the relation between Heb. ‘ēṣ and Aram. “is more plausibly explainable by a formula *‘ḍ > Aram. *“>” than by *‘ḍ > Aram. “, since such a development would have created the effect *wḍ’ > Aram. *w”, while the historical counterpart of Heb. -ws’ in the same type of Aramaic is w”. The Indo-European pendant of the problem under discussion is the question whether in those roots, in which some dialects have two labiovelars while others show a labial and a labiovelar, the latter constellation is the original one and the former the result of assimilation or whether the former represents the original state of affairs and the latter has been created by dissimilation; thus, for the numeral ‘5’: *qenqe (Lat. quinque) > *penqe (OInd. pañca) or, inversely, *penqe > *qenqe? Here, the decision in favour of the dissimilatory solution (contrary to traditional view) can be reached by considering languages in which (as a consequence of the total or partial loss of distinct labiovelar articulations) former labiovelar consonants are, in all cases, represented by the same phonemes as the (original) labials (e.g. Goth, fimf, Extra-Latin Italic pump-) within which, consequently, no assimilatory process can be assumed, or by considering variant locations of the non-labial component (Lat. spec- and Ger. spähen versus Gk. skopós, cf. my Laut- und Formenlehre der herodotischen Sprachform, Heidelberg, [1962], 54Google Scholar) even within the same language (Mycen. Gk. artopóqos versus Ionian and Attic Gk. artokópos ‘bread-cooker, baker’ in view of Lat. *quequo > coquo and OInd. pacati ‘cook’); cf. Lejeune, M., Mémoires de philologie mycénienne, Paris, 1958, 302Google Scholar: ‘II y a eu, en mycénien une tendance … à la dissimilation d'une labiovélaire en labiale sous l'action d'une autre labiovélaire dans la partie suivante du mot’. Analogous statements (mutatis mutandis, according to whether the dissimilation is pro- or regressive) can be given validity also for most other Indo-European languages and, with respect to the emphatics, also for Semitic.
10 Atti del xixmo Congr. Intern, degli Orientcdisti, Roma, 1935, Roma, 1938, 451.Google Scholar
11 This tends to corroborate the reconstruction of two Canaanite ‘ṣ’-phonemes transmitted in the Greek forms Sidón and Týros, although, of course, two other Semitic phonemes are involved here.
12 Arabic ẓ seems to have the same correspondence as ṣ: ṣōhar-zōhar go with ẓuhrun-zahara.
13 The ‘breaking’ notion unites, as a posited common semantic component (Grundbedeutung, if this term may be considered admissible), the significations' break out in wrath' (comparison with the uses of Gk. hrḗgnumi is made in some editions of Gesenius's dictionary) and ‘small broken branches, undershrub’ (qeṣep = LXX phrúganon, Hos. x, 7, qṣåpå = LXX sunklasmós, Joel, i, 7).Google Scholar
14 It is very often overlooked that systemic symmetry as well as etymological affinities require that Canaanite š (rather than s) be regarded as the unvoiced counterpart of z, which leaves s with no voiced counterpart.
15 cf. Gordon, C. H., Ugaritic textbook (Analecta Orientalia, 38), Rome, 1965, 27.Google Scholar
16 See p. 446, n. 12.
17 Biblical ṣrp is a cognate of Ar. ḍrb—as will be demonstrated below; post-Biblical Hebrew ṣrp ‘direct towards, link with’ corresponds to Ar. ṣrf. However, there are some collocations of the latter that might necessitate an inquiry whether there is not an Arabic root ṣrf 11, conveying the notion of ‘purity, unmixedness (such as of wine)’ issuing from ḍrb as a result of a process of unvoicing (cf. above) and approaching the semantic character of Heb. ṣrp, Akk. ṣarāpu.
18 Gk. háptesthai ‘touch’ or ‘be ignited’, háptein ‘ignite’, Ger. anstecken ‘kindle’, etc. (cf. Buck, C. D., Dictionary of selected synonyms in the principal Indo-European languages, Chicago, 1949, 76–7Google Scholar); the Turk, root yak- for ‘getting (in) close (contact)’ and ‘igniting, singeing’; Heb. dlq ‘catch up with … (in a pursuit), be kindled’.
19 cf. Lat. pū-rus and the OInd. verbal root pū- ‘purify’ in relation to the ‘fire’ root extant in Gk. pû-r, fi-re etc.; see Hammerich, L. L., Laryngeal before sonant (Kgl. Daneke Vidensk. Selsk. Hist.-fil. Medd., XXXI, 3), 1948, 36.Google Scholar
20 Ezek. xxi, 3Google Scholar; Prov. xvi, 27.Google Scholar
21 We will not adduce here šåråb, although in Isa. xlix, 10Google Scholar, it is collocated with yak-ēm.
22 ‘Indicium, imperium’ (Mandelkern, S.v.). Cf. the rendering of ḫådlu- pråzo-n b-yiśrå'ēl (Judges, v, 7Google Scholar) by exélipon dunatoì en Israēl in the Codex Vaticanus, while the rest of the LXX tradition leaves the word untranslated (exélipen phrazōn en tôi Israēl). Ṣidqōt p-irzōno- b-yiśrå'ēl (ibid. 11) is paraphrased by díkaioi eníschusan en tôi Israēl (except for the Vatican codex which has dikaiosúnas aúxēson en Israēl). In Jewish exegetical tradition (of. Torczyner's note in Ben-Yehuda's Thesaurus, S.v.), the word has been interpreted on grounds of the post-Biblical p-råzo-t ‘open country’, which seems to offer little benefit for the understanding of the Biblical expressions quoted.
23 Of no apparent relationship with the ‘duality’ root prd; see below.
24 cf. p. 451 n. 29.
25 The relations between the Biblical ‘containing + prohibiting’ root ‘ṣr ‘cogere’ and the Biblical ‘tension’ or ‘string, binding’, post-Biblical ‘prohibiting’ lexeme ‘sr as well as ‘sr ‘coagere’ and their respective Arabic equivalents ‘sr ‘constringere’, ‘dr’ premere, prohibere’, ‘sr ‘ligare, coercere’ may be clarified and sorted out on very similar lines, but cannot be gone into here.
26 The problem of the (yellow-reddish?) colour term Ar. ṣahiba, Heb. ṣåhōb opposite dahabun, zåhåb ‘gold’, which my colleague J. Blau was good enough to bring up in discussion, belongs to a different area of phonology. If the possibility of the ‘gold’ term being a loan-word is considered, one should envisage a Semitic source-language, which would permit us to maintain the etymological link between the two lexemes.
27 It must be borne in mind that Akkadian has no /e/-phoneme (cf. von Soden, , Grundriss (Analecta Orientalia, 33/47), Rome, 1969, § 8b, p. 10Google Scholar, but see the discussion and literature in Reiner, , JCS, xxv, 1, 1973, 45Google Scholar), just as it has no /o/-phoneme. The materialization [e] can, consequently, only be deemed as representing a vowel plus (in Saussurian terms) a ‘coefficient’ which we have accordingly symbolized by a capital letter (p. 443, n. 2). It is the presence versus the absence of that ‘coefficient’ that can produce distinctive oppositions.
28 We must consider cases in which no other potential colouring factor is present; e.g. in eṣēru ‘to give form’ the ‘vocalic’ initial is potentially a colouring factor and it corresponds, in fact, to Heb. yṣr; and likewise the presence of two consonants only in ṣēlu ‘rib’ (Heb. ṣēlå', Ar. ḍil'un).
29 Akk. ṣar7amacr;ḫu ‘light up, glow’ could, in fact, constitute the same word, if we assumed a basic notion of ‘acuity’ materializing either visually or acoustically (cf. the possible relation between Akk. ṣarāru ‘spangle’ and Ar. ‘make a sharp sound’ discussed below); in that case, the Hebrew verb zåråḫ ‘shone’, doubtless a cognate, would show the regular dissimilatory de-emphasization *ṣ > z. Or, if we assume that there were two separate Semitic roots, *ṣrḥ, and *zrḥ (or *drḥ), the creation of root homophony in Akkadian would have been due to a process of assimilatory emphasization, which is, at any rate, not totally unheard of in Akkadian, cf. qaṣāru ‘bind’ in relation to Heb. qåšar ‘idem’ and, if a semantic development parallel to that posited for ‘ṣr–'sr (p. 449, n. 25)Google Scholar is not excluded, also in relation to Ar. qasara ‘induce by force’. The abstract noun ‘brilliancy’, ṣarḫu, has optional e-colouring, which is the rule for ‘giving sound’, ṣerḫu; this is due to the /E/ reflecting *ḥ, which is here what is represented by ḫ (cf. p. 443, n. 3).
30 We cannot enter into a treatment of parṣu ‘commandment, rite’, parāṣu ‘break through’, and the ‘spreading’ root prš (in naprušu ‘fly around’) due to the intricate developments undergone by roots of that structure.
31 For this notation see p. 443, nn. 2–3.
32 But with an isolated ulteṣbit (von Soden, , Grundriss, § 87b, p. 117).Google Scholar
33 There seems to be a polarized contrast between this Akkadian root and weṣû ‘lay out’ (Ar. waḍa'a), in which /E/ = *’.
34 Taken over from Bezold's dictionary?
35 Akk. bêru ‘pit, well’ corresponds to i-base nouns in Arabic and Hebrew, but an additional *a(?)-base noun exists in Hebrew, bōr.
36 For the semantic aspect, of. p. 451, n. 29.