Printer Friendly

The I-w verbal class and the reconstruction of the early Semitic preradical vocalism.


It is a truism of linguistic research that synchronic irregularities - deviations from the general pattern of a given linguistic system - frequently afford insights into patterns no longer extant in the system. In the case of the verbal systems of the various Semitic languages, one of the most consistently irregular subsystems is found in fientive (i.e., non-"stative," from the point of view of morphology(1)) basic-stem verbs whose initial radical is reconstructed as *w. The morphological idiosyncrasies of the I-w class have long been recognized and are so conspicuous and so widespread throughout the Semitic languages that one might be tempted to describe them as deviations from the norm which reach as far back into history as we can reconstruct. In the following pages, however, it will be suggested, not only that, despite their eccentricity, the anomalous structures typical of the I-w class have actually arisen from what was the original regular, strong-verb type, but that the I-w type provides crucial evidence for the reconstruction of the very core of Semitic verbal morphology.


In characterizing the synchronic deviations typical of the I-w fientive class, we must distinguish between the West Semitic and the East Semitic languages. A clear example of the shape assumed by the I-w class in the West Semitic languages may be found in Classical Arabic. The distinctive characteristics of this class in Arabic are (1) the loss of the radical w in the imperative, present, jussive, and subjunctive forms (i.e., the imperative and those forms with a prefixed subject-marker); and (2) the systematic appearance of the thematic vowel i (or a, in the environment of a post-velar radical or [r.sup.2]) in these same forms, at the expense of the other theoretically possible thematic vowels u and a. The past tense stem (wa[C.sub.2]a[C.sub.3]-), unlike the imperative/prefixed stem, follows the regular pattern.(3) In contrast to the irregularities of the I-w fientive class, I-w stative verbs behave like normal verbs of the samia-type or the kabura-type. Contrast the paradigm of Arabic wajada 'he found' with that of a regular fientive Arabic verb such as kataba 'he wrote' (which belongs to the subclass containing u as the thematic vowel) and the paradigm of wabira 'he was hairy' with that of a typical stative-pattern verb, samia 'he heard'.


There is no explanation within Arabic for the replacement of the expected stem-shape -w[C.sub.2]V[C.sub.3]- (where V = any one of the vowels u, a, or i, as stipulated by the lexicon) by -[C.sub.2]i[C.sub.3] in I-w fientive verbs.

The formation of the I-w verbs in Biblical Hebrew is in close agreement with the Classical Arabic pattern. Here too there is a distinction between two basic configurations, corresponding to the Arabic fientive and stative types. The prefixed stem of the stative type takes the basic shape yiCaC- (< yi-yCaC- < yi-wCaC-, i.e., with the first redical retained, like Arabic ya-wbar-u) - e.g., yisan 'he sleeps' (= Arabic ya-wsan-u), wat-tiqad 'and it burned' - showing evidence of the earlier preradical vowel i expected on the basis of Barth's principle (Barth 1894; see III.3. below). The fientive type shows the same predilection for a thematic i-vocalism that Arabic does, as well as the same loss of the first radical - perfective yasab 'he sat' (< wavab-) vs. imperfective yeseb (< yi-vib-),(4) imperative seb (< vib); as in Arabic, there are examples of verbs which vacillate between the two patterns (e.g., imperfective yeqar (yeqar) 'he is valuable' following the fientive pattern, but waw-consecutive way-yiqar with the stative pattern).(5)

The structure of the fientive I-w class in Hebrew may thus be traced back to an earlier pattern comparable to that of Classical Arabic (yeseb < yi-sib- < ya-vib-, like Arabic ya-vib-u). To the extent that its script allows us to determine, Ugaritic appears to agree with this reconstruction - cf. a-bl 'I bring', y-rd 'he goes down', t-ld-n 'they (du.) bore', rd 'go down (imperative)' (Gordon 1965, 85-86; Segert 1984, 70-71). Phoenician also appears to agree in showing the absence of the first radical, e.g., y-sb (Segert 1976, 147). The first radical is also absent in Aramaic - cf. Old Aramaic (Sefire III) y-sb, imper. sb-w, Biblical Aramaic yi-ttib, Palestinian Jewish Aramaic yi-tteb. In Syriac a reflex of the fientive pattern survives in ne-tteb 'he sits' (with secondary gemination of the second radical, seen also in the Biblical and Palestinian Jewish Aramaic forms) and in imperatives such as teb 'sit' and hab 'give', while elsewhere the stative-stem pattern (< yiCaC- < yiwCaC-) has expanded to provide a new general I-w model, cf. ni( )lad 'he begets', ni( )rat 'he inherits', spelled with an extraneous aleph on the model of the I- class (Brockelmann 1981, 92-94).

The two types of I-w verbs are also clearly attested in Geez (Dillman 1907, 180-81; Lambdin 1978, 191-92), although some restructuring has evidently also taken place in this language. The jussive and imperative stems show the familiar loss of the radical w in many instances (jussive ye-lad, imperative lad = Arabic ya-lid, lid; jussive ye-rad, imperative rad = Arabic ya-rid, rid), although the presence of the thematic vowel a rather than e < *i is unexpected. At the same time, the survival of the radical w, as injussive ye-wgar, imperative wegar (alongside ye-gar, gar), is not rare. As in the w-less forms, it is surprising to note that the theme-vowel is the opposite of what we might expect if we compare the Geez forms retaining w with the stative-type of the other West Semitic languages (yi-w[C.sub.2]a[C.sub.3]-) - i.e., yerad rather than yered (< yi-rid-), yewger rather than yewgar (< yi-wgar-). It is quite likely that the distinction between Geez ye-gar and ye-wger results from the general tendency for irregular verbs to drift toward the model of the regular verbs (leading to the appearance of past wagar-a / juss. yewger / pres. ye-wagger < wagar-a / ye-ger / ye-wagger, on the analogy of regular verbs such as nagar-a / yenger / ye-nagger) and therefore has nothing to do with the distinction between fientive ya-vib-u and stative ya-wbar-u in the other West Semitic languages.(6) The other Southwest Semitic languages seem to support the Arabic model - cf. the Epigraphic South Arabian present y-z corresponding to past wz (Hofner 1943, 94), and the Jibbali (Sheri) subjunctive y-red (corresponding to past ?? 'he went down for water') (Johnstone 1981, xviii).

The I-w verbs of the West Semitic languages, in short, attest to an opposition between fientive ya-[C.sub.2]i[C.sub.3]- (with imperative [C.sub.2]i[C.sub.3]-) and stative yi-w[C.sub.2]a[C.sub.3]-. Such an opposition, we have noted above, defies a simple phonological explanation, in that the fientive form ya-[C.sub.2]i[C.sub.3]- cannot readily be derived from the expected ya-w[C.sub.3]V[C.sub.3]- by any phonological process yet identified. It has been suggested (Brockelmann 1908, 596-97; Bauer and Leander 1965, 378-79) that these forms owe their shape to a loss of the initial syllable in the imperative stem (i.e., wivib- < vib-) and a subsequent restructuring of the present stem form on the basis of the imperative (ya-wvib- < ya-vib- on the analogy of vib-). There is no support elsewhere in the Semitic languages, however, for a loss of initial wi-, rendering a putative change of this sort ad hoc. Moreover, it remains difficult to explain the absence of the first radical in Arabic, where the imperative stem would presumably have taken the form iwvib- (which would have gone to ivib-, like ibar- < iwbar-) paralleling idrib, rather than the form witvib- indicated by the other Semitic languages.

Other analyses, in contrast, have preferred to take the truncated form of the stem (vib-) to be the original shape, and the radical w- of the past stem (wavab-) to be a prefixed element by which what was originally a biradical stem has been incorporated into the triradical model of Semitic morphology. For example, Ahrens observes:

... Wenn nun Brockelmann ... behauptet, der Imper. thib (yon wathaba) habe diese Form aus phonetischen Grunden, indem schon im Ursemitischen von der vorauszusetzenden Form withib die Silbe wi "zur Vermeidung der heterogenen Lautfolge" abgeworfen sei, so darf man an dieser ad hoc aufgestellten Lautregel wohl Zweifel hegen; wir wetden vielmehr, da nicht nur das Hebr., sondern auch das Arab. diese Gruppe yon Verben gewisse Formen ohne das Waw (Hebr. Jod) bilden lasst, vermuten durfen, dass jedenfalls fur diese Formen das Waw uberhaupt nicht ursprunglich zum Stamme gehort hat, dass wit vielmehr in den Formen mit Waw (bzw. Jod) eine jungere Entwicklungsstufe zu sehen haben ... (Ahrens 1910, 179)

Such an interpretation of the I-w class begs the question, however, of the nature of the stem-addition wa-, since this explanation would require some measure of independent grammatical motivation before it could be thought of as any less ad hoc than Brockelmann's view.


Let us contrast the West Semitic I-w patterns (fientive -[C.sub.2]i[C.sub.3]- vs. stative -w[C.sub.2]a[C.sub.3]-) with the patterns seen in Akkadian. Here too we find a morphological distinction between fientive and stative classes. The latter class is formally characterized by the necessity of reconstructing the first radical as *y rather than as *w in those elements of the paradigm in which the first radical is preceded by a subject-prefix. The portion of the paradigm containing subject-prefixes thus merges with the pattern of the I-y class.


I have suggested elsewhere (Testen 1992) that the paradigm of the stative I-w class of Akkadian may ultimately be traced back to an original yi-w[C.sub.2]a[C.sub.3]-, which is directly comparable to the West Semitic model, the shift of *w > *y having been caused by$the earlier presence of an adjacent *i-vowel such as Barth reconstructed for G-stem verbs of the stative pattern (e.g., yi-sma-). Mutatis mutandis the Akkadian isimtype may therefore be compared to the pattern of Hebrew yiras and dialectal Arabic yibaru (vs. normative Arabic yawbaru - see Wright 1979, sec. I:79 rem.).

The problem of the Akkadian fientive I-w class is considerably more complex. In these verbs we find that the preradical vowel is consistently u. Compare the paradigm of (w)abalu(m) 'to carry' against the paradigm of a regular verb such as parasu(m) (the thematic vowel of which is u):
Present ubbal iparras
Preterite ubil iprus
Perfect ittabal iptaras
Stative (w)abil paris
Verbal Adjective (w)abl-u(m) pars-u(m)
Imperative bil purus

Since early Semitic aw resulted in Akkadian u, it has been supposed that the preterite ubil is to be reconstructed, not on the model of West Semitic ya-bil-, but as ya-wbil-, i.e., without the loss of the radical *w. It has therefore been concluded that the preradical u of the Akkadian form is to be read as a long vowel (as would be expected of the reflex of an earlier diphthong; yon Soden 1969, 139). Despite the discrepancy between the initial syllables of the Akkadian preterite ubil and the West Semitic ya-vib-, however, it should be noted that Akkadian and the West Semitic languages agree exactly in showing the therustic vowel i, as well as in the form of the imperative stem (Akkadian bil and the West Semitic type vib-).

The interpretation of the Akkadian I-w form as having arisen from yaw[C.sub.2]i[C.sub.3]- is rendered more complicated by the effects of the Akkadian syncope rule. In Old Akkadian, Old Babylonian, and Old Assyrian, we typically find the elision of the vowel of the preterite stem's second syllable in those forms in which that syllable is opened through the addition of a following vowel - 3 sg. ubil, 3 pl. ubl-u, 3 sg. ventive ubl-am (von Soden 1969, 139). Such a vowel loss is without parallel after a long syllable, but entirely normal after a short syllable (cf., e.g., the G-stem participle paris-u(m) vs. the verbal adjective pars-u(m) < paris-um). The loss of the thematic vowel of the I-w preterite therefore entails either (a) an analysis whereby the preradical vowel is to be read as short, and the syncope is therefore regular; or (h) an analysis whereby the preradical vowel is to be read as long, and a new morphophonological rule must therefore be formulated for verbs of this type(7):

(a) ubil- + -am [right arrow] ubl-am by general phonological rule

(b) ubil- + -am [right arrow] ubl-am by special rule

All other things being equal, the first of these two options is to be preferred, since it relieves us of the necessity of positing an ad hoc syncope rule. Indeed, the only motivation for choosing the second option has been based on the preconception that ubil is to be traced back to ya-wbil-, a reconstruction which, we have seen, runs counter to what we find attested in West Semitic. Adopting the reading with short u thus facilitates our interpretation of Akkadian synchronically, but only, it would appear, at the expense of rendering the preradical u of ubil both morphologically anomalous and historically opaque.


A comparison of the Western and Eastern manifestations of I-w verbs thus appears to lead to the following conclusions:


At the same time that we observe the disconcerting disjunction between the West Semitic languages and Akkadian in the formation of the fientive stem, we cannot disregard the striking elements which the two branches share: the consistent presence of the thematic vowel *i (rather than *u or *a, which in principle ought to have been just as possible as *i) and the absence of the first radical in the imperative. The co-occurrence of these idiosyncrasies in both the Eastern and Western branches renders an equation between the ancestor of the Akkadian preterite yu[C.sub.2]i[C.sub.3]- and the ancestor of West Semitic ya[C.sub.2]i[C.sub.3]- a clear desideratum. For the moment, while we must admit that obstacles confronting such an equation appear formidable, we may take note of at least one point which both East and West seem to have in common - the shortness of the first syllable, whether its vowel be traced back to *u or to *a.


Let us temporarily leave aside the specific problem of the I-w class and consider the issue of the Semitic preradical vowel in general. On the basis of the agreement of Arabic and Akkadian,(8) we find that we can reconstruct two basic patterns for the preradical vowel corresponding to two major classes of verbal stems: (1) what we shall call for the moment the *a/*i-type, as seen, for example, in the basic (i.e., G-) stem or N-stem (= Arabic Form I and Form VII, respectively); and (2) the *u-type, seen in, the D-stem or S-stem (= Arabic Form II and Form IV).
 Arabic Akkadian

G-stem/Form I a-fal- a-prus
N-stem/Frm VII a-nfail- a-pparis
D-stem/Form II u-fail u-pparis
S-stem/Form IV u-fil- u-sapris

The *u-class maintained the vowel *u throughout the paradigm, for all verbs belonging to the stems constituting this class. The *a/*i-class, in contrast, displayed clear propensities toward an alternation of *a and *i in the preradical position, but the data which the languages provide concerning this alternation are complicated. We must distinguish between the *a/*i alternation patterns of Akkadian, on the one hand, and of the West Semitic languages on the other.


In Akkadian the alternation between preradical a and i in the *a/*i-type of verb occurs within the paradigm. Consider the preterite paradigm of parasu(m):
 Singular Plural

1 a-prus ni-prus
2 masc. ta-prus ta-prusa
2 fem. ta-prusi
3 masc. i-prus i-prusu
3 fem. ta-prus(9) i-prusa

It has been argued on the basis of comparative Semitic evidence that the third-person marker i- of Akkadian is to be traced back to ya- (von Soden 1969, 23). It is not entirely certain that i- is the expected development of ya- in Akkadian, however, since, aside from the putative case of the third person prefix, ya- seems frequently to result in e- in Assyrian: cf. the infinitive idu(m) (Ass. edu(m)) < yada-um, and the adjective isaru(m) (Ass. es(a)ru(m)) < yasar-um (von Soden 1969, 23). It is therefore difficult to maintain that ya-developed into Assyrian i- specifically in the case of the verbal prefix. In any event, there is no phonological means available to trace the prefix ni- of the first-person plural back to an earlier na-. We must therefore assume that the early East Semitic a-type paradigm featured an alternation between a prefix vowel -a- (in at least the first-person singular a-, the second-person [singular, dual, and plural], and the third-person singular feminine ta-), and -i- (in the first-person plural ni- and, quite probably, in the third person [other than the feminine singular] (y)i-).


The *a/*i alternation reconstructed for early West Semitic is of an entirely different nature. Barth (1894) assembled evidence for the reconstruction of *a as the preradical vowel for G-stem verbs containing *u or *i as thematic vowel (e.g., ya-ktub-), while the reconstructed preradical *i was restricted to verbs with a thematic vowel *a (e.g., yi-sma-); Ginsberg (1939) demonstrated that this alternation was evidently still retained as part of the live morphology of Ugaritic. The West Semitic alternation thus evidently did not occur within the paradigm, as the Akkadian alternation did, but among paradigms according to thematic vowel classes. Subsequently the opposition between *a-preradical verbs and *i-preradical verbs was lost in the majority of the descendent languages, one alternant or another becoming generalized so that each of these languages developed a characteristic reflex of an -a- or -i- preradical (thus Arabic yaktub-, yasma- vs. Geez yekteb, yesma).(10)

Following Barth, we therefore reconstruct the ancestral vocalic pattern of the G-stem verbs of the West Semitic languages as shown in the following table.(11)


The *a/*i alternation reconstructible for the West Semitic languages is thus substantially different from that of Akkadian in that any given verb retained the same preradical vowel throughout its paradigm. However, if my interpretation of the I-w verbs of the Akkadian stative type (Testen 1992) is correct and the anomalous paradigm of isim is to be ascribed to the influence of Barth's Law, we may reconstruct an *a/*i alternation of the West Semitic sort also for an early stage of East Semitic. The distinction in the preradical vowels associated with *u- and *i-thematic stems and the preradical vowel with *a-thematic stems may thus be assigned with considerable confidence to early common Semitic, as Barth himself proposed in 1894.


How then are the *a/*i-pattern identified by Barth and the paradigm-internal alternation found in Akkadian to be reconciled? The key to the Akkadian alternation, I suggest, lies in the nature of the consonant which constitutes the person-marker: for the person-markers - and *t- the preradical vowel was -a- (> aprus, taprus), while for the person-markers *y- and n- the preradical vowel was -i- (>iprus, niprus). We may therefore characterize this alternation by stating that -i- is found in conjunction with sonorant subject-markers, whereas -a- is found in conjunction with non-sonorant subject-markers.

Let us now consider the nature of the *a/*i opposition underlying Barth's theory of the Semitic preradical. It is worth noting that Barth framed his proposal simply in terms of the observed distribution, rather than as an attempt to explain how such an opposition could have arisen (for example, through dissimilation). There is in fact no non-ad hoc means available to derive the elements of this alternation from either original Semitic *a or *i. Thus, taking the *a-preradical forms as the starting point, we find it impossible to identify any known Semitic phonological phenomenon which could have caused an original *a to become *i under the influence of an *a in a following syllable (ya-sma- > yi-sma-); and, on the other hand, it is equally difficult to explain the *a-preradical forms on the basis of the *i preradical (yi-ktub > ya-ktub-).

The appearance of a vocalic sequence -i- ... -a- is, however, by no means unfamiliar in Semitic. Consider the shape assumed in the various languages by the imperative of a G-stem verb whose thematic vowel is -a-:


These forms may be traced back to rikab-, sima- in Northwest Semitic and Geez. The Arabic imperative, by contrast, shows (i)rkab-, (i)sma-, in which a vowel i is located, not after the first radical, but before it. The Akkadian imperative, finally, shows two vocalic patterns with a-stem verbs, rikab and sabat, the first of which, judging on the basis of pre-Hebrew, etc., rikab-, is in all likelihood the original.(12)

The difference between the syllabic configuration of Arabic (i)rkab and that of pre-Hebrew rikab notwithstanding, the appearance of the vowel pattern *i ... a throughout the Semitic languages is striking. Imperative stems for G-stem verbs appear to have been formed in early Semitic through the introduction into the preterite/jussive stem (either through infixation or, for Arabic, prefixation) of a high vowel agreeing in roundedness with the vowel of the stem.(13)


It is important to note that the vowels in question (i.e., the vowels located either before or after the first radical) have, to all appearances, been introduced secondarily. The G-stem imperative is best thought of as having had the original structure *[C.sub.1][C.sub.2]V[C.sub.3]- (i.e., the structure of the prefixed conjugation stem without the person-prefixes). This reconstruction of the imperative stem is supported by Classical Arabic, where the secondary nature of the prefixed vowel (i)/(u) is transparent. The prefixed syllable of the Arabic imperative appears only in the absence of a preceding vowel. Consequently, we find that, when necessary, the underlying word-initial consonant cluster *[C.sub.1][C.sub.2]- receives a prothetic syllable (i) or (u), by which the language's restriction against such clusters occurring at the surface level is maintained (isma 'hear' but wa-sma 'and hear').

It therefore appears that the vocalic sequence *i ... a in the imperative was the result of the introduction of an intrusive, historically secondary vowel *i into a consonantal complex which would otherwise have been syllabically disallowed. May we not conjecture that the sequence *i ... a seen in Barth's view of the preradical and thematic vowels owes its appearance to a similar phenomenon? If we speculate that in early Semitic the prefixed conjugation of the G-stem simply consisted of the prefixing of the consonantal subject-markers directly onto a verbal stem having the form [C.sub.1][C.sub.2]V[C.sub.3] (i.e., the shape of the bare imperative), we can readily imagine that the introduction of an epenthetic vowel to disrupt the resulting triconsonantal cluster would have been all but unavoidable.
1 sg. -ktub-
1 pl. n-ktub-
2 (sg., du., pl.) + 3 t-ktub-
3 (other than y-ktub-

It is in this connection that the a/i alternation seen within the Akkadian paradigm assumes particular significance. As we have remarked above, this alternation appears to be predicated upon the phonetic nature of the person-marking consonant - the preradical vowel i is associated with sonorant subject-markers and the preradical a with non-sonorant subject-markers. This distribution may be related directly to the hypothesis sketched above if one posits that the non-sonorant consonants were incapable of appearing in clusters of the sort described above without the introduction of a vowel. Sonorants, on the other hand, being capable themselves of constituting a syllable core, had no need for an intrusive vowel. I therefore conjecture that the schematic paradigm given above took the following form (where indicates the inserted vowel and a superscript dot indicates a sonorant serving as a syllable-core):
1 sg. -ktub-
1 pl. n-ktub-
2 (sg., du., pl.) + 3 t-ktub-
3 (other than y-ktub-

The *a/*i opposition identified by Barth, I suggest, results from the development of the intrusive vowel into *i before an *a in the following syllable (i.e., first-person singular sma- > isma-, just as the imperative sma-/sma- > sima-/(i)sma-) but into *a before a high vowel (ktub- > aktub-).(14) Since the underlying vowel is ultimately neither Semitic *a nor *i, one obviates the difficulty of explaining how Barth's distribution could have arisen, since neither of the original Semitic vowels *i or *a shows the dissimilative properties called for by the *a/*i opposition.

For the sonorant person-markers y- and n-, I hypothesize that the person-markers themselves served as the syllable, rendering the introduction of an epenthetic vowel *v unnecessary. This phenomenon, I suggest, is reflected directly in Akkadian i-prus, the i- of which I take to be simply the syllabified counterpart of y-. It is not beyond comprehension that a syllabified subject-marker n- could have taken the form ni- in Akkadian.(15)

In short, the preradical system of early Semitic may be reconstructed as consisting, not of an opposition of -u- vs. -a/i-, but of an opposition between a preradical u- (cf. the D-stem -u-qattil-, y-u-qattil- and the S-stem -u-(s/h/0)a-qtil-, y-u-(s/h/0)a-qtil-) and a preradical -0- (cf. the G-stem -0-qtul-, y-0-qtul- > a-qtul, i-qtul-). This is what we find, in fact, attested in Akkadian, assuming only the loss of Barth's preradical *i (sma- > isma- > asma-, ultimately > Bab. esme). The remaining Semitic languages have not only generalized one of Barth's alternants over the other (as Barth proposed) but have also erased the alternation within the paradigm which we still find in Akkadian. This has been achieved by expanding the -a- or -i(<v) originally associated with the non-sonorant person-markers to the sonorant person-markers, thus creating the new prefixes ya-ktub- (or yi-ktub-) and na-ktub- (or ni-ktub-) to replace original *i (i.e., "*y")- and n-. As a result of these simplifications, we find a symmetrical set of person-marking prefixes a-, ta-, ya-, na- (or i-, ti-, yi-, ni-) paralleling u-, tu-, yu-, nu-, such as we find attested in Classical Arabic.


A subsequent stage in this development may be seen in those West Semitic languages in which the preradical vowel *u has been lost, and the opposition between the two original types of preradical vocalization consequently eliminated.(16) In a number of West Semitic languages, moreover, a new preradical -u- has appeared (either through a reemployment of the original *u or through the introduction of a new *u) in connection with the development of an ablaut-passive (Hebrew yopal, Arabic yufal-, etc.).(17)


I have hypothesized, therefore, that the "*a/*i" - class preradical vowel is ultimately to be traced back to a phonological intrusion appearing within initial clusters of the sort described above, i.e., those in which the elements of the cluster itself were incapable of constituting a syllable. It would appear entirely likely, however, that in clusters of this sort syllabicity could be assigned, not only to the person-markers n- and y-, but also to the stem radical [C.sub.1] which formed the core portions of such clusters, assuming that the phonological nature of the [C.sub.1] in question was such that syllabicity was an option. It is in this light that we must go back and consider the I-w type.

If one conjectures that the early Semitic phoneme *w was capable, like the person-marker *y, of taking syllabicity upon itself and shifting from semivowel to vowel, one finds that the addition of the person-markers would lead to the following paradigm:


This is substantially the situation as it is found in Akkadian - the first radical *w has (to all appearances) vanished and a short vowel -u- has appeared between the subject-marker and the stem. The only discrepancy between this model and the Akkadian paradigm as it is attested is the restriction of the thematic vowel to -i- rather than to *a, *u, *i. This phenomenon finds a ready explanation, however, once we note that -i- is the characteristic thematic vowel of verbs of the u-preradical type (Akkadian uparris, Arabic yuqattil-, Hebrew yaqattel; Akkadian usapris, Arabic yuqtil-, Hebrew yaqtil). Given the close resemblance of the initial syllable of the I-w type as we reconstruct it here (u-, tu-, yu-, nu-) to the initial syllable of the u-preradical verbal types, it is not surprising to find that the thematic vowel of the latter class of verbs has been borrowed as the characteristic theme-vowel of the I-w verb (-u[C.sub.2]i[C.sub.3]-).

It is suggested, therefore, that in East Semitic we find the clearest reflection of the original shape of the I-w paradigm: an initial syllable with short *u and, in connection with this, a secondary *i as the thematic vowel. East Semitic reveals, moreover, a stage through which West Semitic must have passed at some point previously. The West Semitic paradigm of the type of Arabic ya-lid-u represents a further development along the same path, consisting of the replacement of the "preradical" vowel *u of yu[C.sub.2]i[C.sub.3]- with the "correct" form of the preradical syllable of a G-stem verb, i.e., the preradical vowel a (ultimately < *v) which was appropriate for a verb whose past-tense form was walad- (i.e., walad-/-u-lid- > walad-/-a-lid-, like darab-/-a-drib-). In the otherwise inexplicable appearance of the characteristic thematic vowel -i-, however, we find a clear indication of an earlier stage in the I-w verb of West Semitic directly comparable to the I-w verb of East Semitic.


From the point of view of the history of Semitic grammar, therefore, the peculiar morphological behavior of the Semitic I-w type (in both its East Semitic and West Semitic manifestations) represents not a primal aberration from the norm but rather a reflection (albeit somewhat obscured by the accretion of subsequent developments) of the original Semitic pattern. Once it has been carefully analyzed, this verbal class enables us to find clear support for what has been posited above - the historically secondary nature of the vowel of the "*a/*i"-preradical type.

1 The "fientive" class is characterized, in Arabic, by the presence of a thematic vowel a in the past tense stem (kataba/yaktubu 'he wrote/writes', daraba/yadribu 'he beat/beats', faala/yafalu 'he did/does'), whereas the past tense stem of the "stative" class contains i or u (wabira/yawbaru 'he was/is hairy', kabura/yakburu 'he was/is great'). In semantic terms, the members of the stative class are predominantly associated with the description of states or characteristics. The vocalic distinctions are not as clear in Akkadian, but in a number of places the morphology of Akkadian makes a formal distinction between semantically stative and semantically non-stative verb-classes - see infra the distinction between usib 'he sat down' and the Zustandsverb isim 'he belonged'.

2 Since the appearance of a vowel a in the place of an otherwise expected high vowel, as is seen in such verbs as ya-da-u 'he lets alone' and ya-qa-u 'he falls', is a familiar phenomenon in the environment of an adjacent pharyngeal or laryngeal consonant, we regard verbs of this type as a subset of the CiC- class and shall henceforth refer to the i-vocalism as a general characteristic of the fientive I-w class.

3 Eight verbs, however, have a past-tense stem with i as the theme-vowel - e.g., waviq-a / ya-viq-u he trusted/trusts' and waliy-a /ya-li 'he was/is near' (Wright 1979, sec. I:78). These might be better thought of as stative-pattern verbs whose prefixed/imperative forms have been restructured on the analogy of the fientive type. A few verbs allow the prefixed/imperative stem to be formed following either the fientive or the stative model - e.g., wahil-a / ya-hil-u or ya-whal-u 'he was/is cowardly'.

4 The absence of reduction in the vowel of the prefix (ye-seb rather than ye-seb) is counter to expectations. It is presumably due either to an irregular retention of the preradical vowel under the analogical influence of the triradical verbs or to a late restructuring of the imperfective (yeseb < ya-ysib) on the basis of the perfective stem, which, as a result of the shift *w- > y-, came to have the appearance of having an initial radical y- (Bauer and Leander 1965, 378-79).

5 It might be speculated that the action of Philippi's Law contributed to the blurting of the two classes by obscuring the distinction between the thematic vowels *i and *a in certain environments in early Hebrew. For example, the unexpected plene spelling of the first syllable of 3 fem. pl. tesabna (vs. 3 masc. sg. yeseb (Ezek. 35:9) looks as if it might represent an earlier stative-pattern, tisabna. This form could indicate a re-structuring of the preradical syllable motivated by the quality of the theme-vowel a, despite the fact that the a was historically secondary in this position (tisibna > tisabna > tisabna [like yiras] > [on the analogy of yeseb, etc.] tesabna).

6 One cannot rule out the possibility that some of the peculiarities of the two types of I-w verbs in Geez are due to phonological developments specific to Ethiopic. Dillman (1907, 181) ascribes the presence of the unexpected a-vowel in yerad to a "strengthening" of *e to a. While this explanation is speculative, it is worth noting that Geez appears to show few if any clear examples of words of the form CeCeC- (i.e., reflecting Semitic CiCiC-, CuCuC-, CuCiC-, CiCuC-). One of the places in which one might expect, on comparative Semitic grounds, to find words of the shape *CeCeC- (aside from the prefixed I-w type yi-rid- mentioned above) is the well-documented broken plural pattern CuCuC- (cf. Arabic kutub-un pl. of kitab-un 'book', sufun-un pl. of safinat-un 'ship', tujur-un pl. of tajir-un 'merchant', etc. - Wright 1979, sec. I:200-203). A plural pattern CeCeC- does not appear in Geez, but we do find the common pattern CeCaC- (e.g., ezan pl. of ezn 'ear', hezab pl. of hezb 'tribe', etc. - Dillman 1907, 301-2). Since the Arabic patterns CuCaC- and CiCaC-, which one might expect on phonological grounds to correspond to Geez plurals of the type CeCaC-, are typically limited to the formation of plurals produced by the truncation of a feminine suffix (e.g., umam-un pl. of ummat-un 'nation', kuwar-un pl. of kurat-un 'district', kubar-un pl. of kubra 'largest (f.)', etc.: Wright 1979, sec. I:199), it might be asked whether Geez CeCaC- plurals might just as easily be thought of as correspondents to the Arabic plurals with the shape CuCuC-, with a "strengthening" of CeCeC > CeCaC of the sort described by Dillman.

7 The latter option, with an irregular syncope following a long vowel, has typically been preferred by investigators (see Greenstein 1984, 36: "... what appears to have happened is that through casual speech or some other cause the vowel /i/ began to elide exceptionally ..."), and the literature cited there.

8 See also Testen (1993a) for discussion of the Modern South Arabian languages Jibbali and Socotri.

9 Distinct from the 3 masc. sg. in Old Akkadian and Assyrian; generally replaced by iprus in Babylonian.

10 Although Classical Arabic generalized the a-preradical, it retained a trace of the earlier i-preradical in ixalu 'I believe'. In early Arabic dialects the -i- was widespread, either following the original pattern described by Barth or generalized in the place of classical -a-. It appears that at least certain of the early Arabic dialects which showed the i- preradical limited this vowel to the subject-prefixes other than y- (iktub-, tiktub-but yaktub-) (Wright 1979, sec. 1:60).

11 We may presume that Barth's role also applied to derived-stem verbs. Since the first vowel of the stems of such verbs is routinely *a, we should probably reconstruct here a preradical vowel *i (e.g., N-stem yi-nkatib-, Gt-stem yi-tkatib- / yi-ktatib-, etc.). In the case of certain of the derived verbal stems, however, the data are inconclusive - cf. Akkadian St u-stapris- vs. Arabic (Form X) ya-stafil-, Akkadian Dt u-ptarris- vs. Arabic (Form V) ya-tafaal-. See also Testen 1993a, 449-50.

12 The fact that the CiCaC imperative pattern of Akkadian is limited to a small set of verbs (e.g., limad, rikab, pisah, tikal), whereas the majority of a-stem verbs form their imperative like sabat, also suggests that the forms with the vowel i represent fossils of an original system. The newer forms in CaCaC have been fashioned on the basis of the productive imperative-formation reflected in the u-stem (purus) and the i-stem (piqid) classes (von Soden 1969, 113). Note that the original a-stem nature of the CiCaC- imperative verbs is supported by the thematic vowel a in their cognates elsewhere in Semitic (Hebrew yirkab = Akkadian irkab; Hebrew yilmad = Akkadian ilmad). By contrast, Akkadian sabat at least has an i-stem cognate (Arabic yadbitu), which opens up the possibility that the a-vowel of isbat (and quite possibly other a-stem verbs) is secondary.

13 This is the case, of course, only for those verb-stems (the majority) in which the presence of two consonants made the vowel epenthesis necessary. Where the stem began with a single consonant (e.g., qum 'rise'), no vowel was added.

14 Note that, outside of the environment of an *a in the following syllable, one must evidently distinguish between the reflex of *v in initial biconsonantal clusters (ktub/(v)ktub > kutub/(u)ktub) and that of *v in initial triconsonantal clusters (tktub- > taktub-). Identifying the details of the reflexes of the posited *v in all its possible environments will, of course, require far more investigation than is possible here.

15 For phenomena directly relevant to the issue of n-, see Testen 1993b.

16 Cf. Ugaritic ashlk (= asahlik-) 'I cause to flow' vs. Akkadian usapris.

17 See also Testen 1993a.


Ahrens, Karl. 1910. Der Stamm der schwachen Verba in den semitischen Sprachen. Zeitschrift der Deutschen Morgenlandischen Gesellschaft 64:161-94.

Barth, J. 1894. Zur vergleichenden semitischen Grammatik. ZDMG 48:4-6.

Bauer, H. and P. Leander. 1965. Historische Grammatik der hebraischen Sprache. Hildesheim: Georg Olms Verlagsbuchhandlung.

Brockelmann, Carl. 1908. Grundriss der vergleichenden Grammatik der semitischen Sprachen. Berlin: Verlag von Reuther & Reichard.

-----. 1981. Syrische Grammatik, 13. Auflage. Leipzig: VEB Verlag Enzyklopadie.

Dillman, August. 1907. Ethiopic Grammar. Tr. James A. Crichton. London: Wilkins and Norgate.

Ginsberg, H. L. 1939. Two Religious Borrowings in Ugaritic Literature. Orientalia, n.s., 8:318-22.

Gordon, Cyrus. 1965. Ugaritic Textbook. Rome: Pontificium Institutum Biblicum.

Greenstein, Edward L. 1984. The Phonology of Akkadian Syllable Structure. Afroasiatic Linguistics 9.1:1-71.

Hofner, Maria. 1943. Altsudarabische Grammatik. Leipzig: Otto Harrassowitz.

Johnstone, T. M. 1981. Jibbali Lexicon. Oxford: Oxford Univ. Press.

Lambdin, Thomas O. 1978. Introduction to Classical Ethiopic (Geez). Missoula, Mont.: Scholars Press.

Segert, Stanislav. 1976. A Grammar of Phoenician and Punic. Munich: Verlag C. H. Beck.

-----. 1984. A Basic Grammar of the Ugaritic Language. Berkeley and Los Angeles: Univ. of California Press.

von Soden, Wolfram. 1969. Grundriss der akkadischen Grammatik. Rome: Pontificium Institutum Biblicum.

Testen, David. 1992. A Trace of Barth's Preradical *i in Akkadian. Journal of Near Eastern Studies 51:131-33.

-----. 1993a. The Loss of the Person-Marker t- in Jibbali and Socotri. Bulletin of the School of Oriental and African Studies 55.3:445-50.

-----. 1993b. The East Semitic Precative Paradigm. Journal of Semitic Studies 38:1-13.

Wright, William. 1979. A Grammar of the Arabic Language, 3rd ed., rpt. Cambridge: Cambridge Univ. Press.
COPYRIGHT 1994 American Oriental Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1994 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Testen, David
Publication:The Journal of the American Oriental Society
Date:Jul 1, 1994
Previous Article:The art of lamentation in the works of Pan Yue: "Mourning the Eternally Departed."
Next Article:A new volume of texts from Hellenistic Uruk.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |