Printer Friendly

The biblical shibboleth story in the light of late Egyptian perceptions of Semitic sibilants: reconciling divergent views.


There are currently three treatments of the phonetics of the shibboleth incident of Judges 12:6 (1) deserving of serious consideration. Hendel (1996) and Faber (1992) present attractive arguments that do not depend on problematic assumptions of differences in the number of sibilant phonemes preserved in the two dialects in question, (2) as so many other explanations of the story have done, (3) including Rendsburg (1988a; 1988b; 1996: 511; 1997: 69f.), whose theory, though ultimately untenable, (4) nevertheless poses a question deserving of an answer (2.5 below).

Both Hendel and Faber rely on no more than a difference of pronunciation of the same inherited Proto-Semitic (PS) phoneme, namely traditional */s/, sin or [S.sub.1], both make other substantive contributions connected with the problem, and although they come to diametrically opposed conclusions regarding the phonetic details, a comparison of the two reveals worthwhile possibilities of reconciliation and harmonization of the best each has to offer. We begin with this comparison.


2.1. Hendel believes that the Gileadite sin was deceptively similar to the Ephraimite samekh. The Ephraimites cleverly substituted samekh for sin in order to pass the test, but not cleverly enough for the Gileadites, who could tell the difference.

Hendel supports his scenario by reinterpreting a parallel, first developed along some-what different lines by Rendsburg (1998a: 257f.; see 2.5 below), between the shibboleth case and a similar Cisjordanian misinterpretation of a Transjordanian name, namely the Masoretic spelling ba'alis (Jeremiah 40:14)--with final samekh--of a sixth-century royal name that appears on an Ammonite seal impression written b'lys'--with sin corresponding to the samekh of the Masoretic text. As Herr (1985: 172) observed, the erstwhile presence in the Hebrew text of the stem-final 'ayin of the name, which is clearly lost in the Masoretic text, is vouched for by the final -a of the Septuagint form belisa (Jeremiah 47:14).

Although one could cavil that Ammonite phonetics need not necessarily be the same as Gileadite, it is hard to regard this parallel as other than convincing despite the interval of around five centuries separating the shibboleth incident from the reign of the Ammonite king. The principle that stable differences of pronunciation between mutually intelligible forms of speech can remain intact over several centuries is demonstrable by reference, e.g., to the difference between Northern British [[??]] and Southern British [[??]] in the stressed syllable of English words like bucket, mother, etc.--which has remained stable for the best part of four centuries (cf., e.g., Dobson 1957: 552 note 6). Moreover the precise phonetics of the situation in each dialect need not have remained totally static for the whole five centuries (see n. 13 below).

2.2. Faber proposes that the sound change that converted the original narrow-groove PS phoneme sin or [S.sub.1] into wide-groove /s/ in various Semitic languages (5) was underway in Hebrew at the time of the shibboleth incident in such a way that the Gileadites already had the wide-groove sin /s/ while the Ephraimites still had the narrow-groove version. According as they were oblivious to or aware of this difference, the Ephraimites either (a) persisted doggedly with their narrow-groove sound or (b) substituted for it [S.sub.2] /s/, either/both of which is/are represented as samekh in the extant version of the tale.

Against Faber's second alternative (substitution of /s/) is the fact that there is no basis for believing that the recorded interpretation of the substituted sound as samekh originated with anyone other than the original, surviving witnesses to the events, i.e., the Gileadites. The old argument (cf., e.g., Marcus 1941: 149; 1942) that samekh in the text is simply a makeshift to overcome the lack of a distinctive symbol for /s/ (or indeed /t/, as required by Rendsburg and others) is not convincing because there are other ways of indicating differences not signaled by an orthography (6) and, prior to the later coalescence of samekh and sin, there is no obvious reason why samekh should have been used to represent a sound that was not samekh. Under this hypothesis therefore the Gileadites would have to have heard the substituted Ephraimite sin as samekh, which implies an unwarranted and unnecessary assumption about the similarity of Gileadite samekh and Ephraimite sin.

Faber appears to support her basic scenario with an assertion (1992: 2) that "if sbolt underwent anomalous developments, it must have been in Gileadite, since that is the only way that the Gileadite initial sound in sbolt would have been absent from the Ephraimite phonological repertory. Alternatively, divergent phonological developments affecting the initial sound in sbolt could have given rise to a sound in Gileadite that was totally absent in Ephraimite." The alternative offered here seems to be between anomalous developments limited to the test word in Gileadite and divergent phonological developments affecting only the initial sound of the test word in Gileadite, in other words perhaps one of Blau's (1977) "weak" (i.e., anomalous) phonological changes versus a more regular set of changes affecting Gileadite generally. It is just possible that the expression "divergent developments" conceals a reference to possible changes in Ephraimite as well, but the use of the plural for the "anomalous developments" (7) which are explicitly restricted to Gileadite leaves this open to doubt. This doubt seems to be confirmed by Faber's eventual decision (1992: 8f.) to regard Ephraimite as conservative while holding the critical change on which the story is based to be a general phonological development within Gileadite. This change is not anomalous but is certainly divergent with respect to what is held to be happening, or rather not happening, in Ephraimite. It is perhaps worth emphasizing that, even adopting the direction of Faber's interpretation, the divide need not be between a dialect with and a "dialect without s." Both dialects, after all, are still held to have possessed the independent phoneme sin. It is only necessary in Faber's terms that the Gileadites perceive the Ephraimite version of this phoneme as (a) audibly different from their own and (b) having sufficient auditory similarity to the Gileadite samekh to satisfy the recorded details of the tale. While this scenario may seem to yield the most straightforward interpretation of the biblical story, Hendel's work has shown that it is not binding.

What Faber has neglected here is the possibility that phonetic change may involve not just the acquisition of a new phone, but just as easily, the loss of an old one, even though her earlier work has repeatedly drawn attention to this latter possibility. (8) It is this neglect that makes it difficult to avoid the conclusion that Faber has provided no real argument in favor of her particular version of events as opposed to Hendel's.

Can a defense of Faber's view of shibboleth phonetics be based on the fact that Egyptian transcriptions of samekh as an affricate appear to continue down to the Twenty-fifth Dynasty (9) (eighth to seventh centuries B.C.)? This, after all, might well suggest that the Cisjordanian Ephraimites substituted affricate samekh for Gileadite fricative sin, yet the Transjordanian Gileadites must have heard this as their own fricative samekh if they imagined that this sound could be substituted for another fricative. Hence, if the Gileadites had a more advanced kind of samekh than the Ephraimites, they would surely have had a more advanced kind of sin as well? Indeed they might; but Faber's (1982: 90) demonstration that fricatives can be perceived as affricates (and, by implication, vice versa), which is so necessary to the first two premises of this argument, just as surely destroys its basic conclusion. In other words, the question of who had fricative and who had affricate samekh is of absolutely no importance to the interpretation of the story and therefore cannot be used to support Faber's thesis (or, for that matter, to attack it). (10)

We could argue this point in more detail using the same four-way opposition between /s/- and /s/-sounds as Faber has employed elsewhere (e.g., 1982: 82f.), i.e., based on the contrast of apical s with laminal s and wide-groove s versus narrow-groove s (and similarly for the affricates c, c, c, and c). We could set up two hypothetical dialects--a more conservative dialect A with, say, sin /s/ and samekh /c/ (11) and a more progressive dialect B with, say, sin /s/ and samekh /s/--and show that whichever way we assign A and B to the two dialects involved in our story we get much the same result, (12) one assignment corresponding to Faber's interpretation, the other to Hendel's. (13)

We are forced to conclude that no convincing argument has as yet been put forward in support of the detail of Faber's interpretation.

2.3. By contrast, apart from the convincing Ammonite parallel (2.1 above and 2.5 below), there are two other points that lend further support to Hendel.

2.3.1. The first of these is the more telling because Hendel could not have adduced it or even been aware of it himself. This is because it depends on Faber's reconstruction of Semitic sibilants, which Hendel misunderstands and therefore rejects (2.4 below), since he has read no more than Faber's 1984 and 1992 papers (despite explicit warnings in both, cf. 1984: 189 and 1992: 8). Intriguing it is, then, that in terms of Faber's theory, Hendel's interpretation, with its more s-like Gileadite sin, i.e., the more archaic form according to Faber, agrees with the suggestion of Gileadite conservatism put forward long ago by Speiser (1942: 12f.) on the basis of cultural differences and dialect geography.

2.3.2. The other point in favor of Hendel's version of events is based on the position that if the Gileadites were sufficiently familiar with Ephraimite speech to devise a successful test, it is likely that the Ephraimites in their turn would have had sufficient contact with Gileadite speech not to be caught totally napping by any test that involved some gross and readily detectable difference of pronunciation. (14) Both Hendel and Faber--in her alternative (b) (2.2 above)--envisage just such a gross difference and both essentially have the Ephraimites thinking they can cope with it by mimicking the Gileadite pronunciation, but alas, the poor Ephraimites inevitably fall down over a subtlety they could not ordinarily have been aware of without special tuition. On this basis Hendel's version is clearly superior because it relies on Gileadite perception of the substituted sibilant simply as samekh, exactly as in the biblical text, whereas Faber's second alternative relies on the unsatisfactory makeshift theory.

This point is to some extent equivocal, however, because of Faber's first alternative, which allows that the entire difference between the Gileadite and Ephraimite pronunciations was too subtle for the Ephraimites even to detect (an alternative to this is that they heard a difference but were too stupid to try to imitate it). In general terms this is clearly a valid position, but it does fly in the face of the Ammonite parallel: if Cisjordanians regularly heard Transjordanian sin as identical to their own (even though the Transjordanians were aware of a difference) and they maintained a distinction between sin and samekh, then there was no basis for them to substitute their samekh for Transjordanian sin. It is of course possible to proclaim the differing bits of evidence dialectally and/or diachronically unrelated. But laudable as such caution or open-mindedness may be, unless there is evidence to back it up, it does not make for a very attractive hypothesis. Other evidence to be discussed below makes it, I believe, still less attractive.

2.4. Hendel's misunderstanding of Faber's epoch-making contribution to our understanding of the history of Semitic sibilants (cf. 2.3.1 above) represents a critical deficiency in the equipment Hendel brings to bear on the problem. It results not merely in Hendel's needless and incorrect claim about Faber relying on a phonemic merger like Blau (1977): (15) it actually prevents him from making an appropriate assessment of the significance of his solution. The fact is that by incorporating Faber's theory of the sibilants into Hendel's explanation, it becomes possible to ground the latter in the wider Semitic context, in which sin during the second millennium B.C. was not a stable or uniform entity.

But that is not quite the end of the story.

2.5. As has been indicated above (2.1), priority in observing the parallelism between the b'lys' case and the shibboleth story (2.1 above) belongs not to Hendel but to Rendsburg (1988a: 257f.; 1988b). In his later paper Rendsburg (1988b: 73f.) adduces another biblical, i.e., Cisjordanian, record of an Ammonite royal name, namely nahas (1 Samuel 11:1), a king dating to the 900s B.C., together with two Neo-Assyrian records all purportedly showing that original Ammonite /s/ was indeed perceived by outsiders as /s/, and thus was in contrast with the <s> of ba'alis. Rendsburg concludes that the Cisjordanian representations with <s> point to Transjordanian retained */t/, since the latter is reliably reconstructed for the PS antecedent of the sibilant of b'lys'. Accordingly Rendsburg reconstructs initial */t/ in the Gileadite test word, (16) an idea that goes back, without anything like such a persuasive argument, via Swiggers (1981) to Speiser (1942; 1967).

Rendsburg's appeal to this additional Hebrew record certainly does much to repair the apparently flawed logic of his original (1988a) reliance on b'lys'. The same is not true, I think, of the Assyrian material, since Assyrian has /s/ ~ /s/ problems of its own. And while Rendsburg provides examples and literature showing that Neo-Assyrian <s> corresponds regularly to West Semitic (17) /s/, he does not do the same for the Neo-Assyrian treatment of the putative Ammonite /t/. In the Tell Fakhariyeh inscriptions Old Aramaic /t/, written <s> in the Aramaic version of the dedicator's name hdys'y (ll. 1, 6, 12) and elsewhere, is represented in the Neo-Assyrian text (ll. 11, 22) as a syllable-final dental stop, which some transcribe as <d> (e.g., Abou-Assaf 1981: 21; Greenfield/Shaffer 1983: 113) and some as <t> (e.g., Millard 1983: 104). (18) This Neo-Assyrian rendition of Aramaic /t/ as anything but <s> may seem to clinch the matter, but Aramaic is not Ammonite. To be totally convinced, we need to have a Neo-Assyrian record of an Ammonite name in which putative /t/ is similarly distinguished from /s/.

At all events, the Hebrew spelling nahas certainly demands an explanation, yet neither Hendel (1996) nor Faber (1992) has a word to say on the matter. (19)

Rendsburg himself (1988b: 74) circumvents one obvious alternative explanation by dismissing as ad hoc the assumption of a sound change /s/ > /s/ occurring between the date of his last Assyrian example (ca. 735 B.C.) and the reign of B'lys' (surely an alternative worth considering). Yet there is an alternative that is not ad hoc and does not depend on etymological niceties, since it applies in addition to another biblical record of an Ammonite royal name, namely sobi (2 Samuel 17:27), which Rendsburg (1988b: 76 n. 14) mentions in order to discard precisely because of its unclear etymology. It is now time to examine this alternative.


3.1. A feature common to the two Ammonite names nahas and sobi (2.5 above) is that the vowel adjacent to the consonant in question in each case is non-high. This contrasts with biblical ba'alis and Ephraimite sibbolet, both of which have a high vowel adjacent to the sibilant. In the case of sibbolet, this high vowel may have been /u/ rather than /i/ during the period in question (cf. Speiser 1942: 10; Rendsburg 1988b: 75; Hendel 1996: 72 n. 1). It is therefore probable that these high vowels conditioned the perception of Transjordanian sin as /s/ in these two words, since there are other phenomena in which similar Semitic sibilants are similarly affected by the same high vowels.

One of these other phenomena is the merger of /t/ and /s/ = /s/ in Old Akkadian, a process that appears to have begun in the vicinity the high vowels /i/ and /u/ (Sommerfeld in von Soden 1995: 36). (20) Another concerns the Late Egyptian perception of Semitic /s/ as /s/ rather than as /s/, a matter now deserving more attention. (21)

3.2. A pilot study of Late Egyptian representations of selected Semitic sibilants (Wood-house 2002: 174-78), based on a relatively uncritical examination of the material collected and interpreted by Hoch (1994: 17-396), supports the hypothesis that although the Egyptian scribes were generally uniform in their phonemic interpretation of a given lexeme (n. 10 above), there was nevertheless a fairly even-handed vacillation between <s> and <s> in their representation of Semitic */s/ (not necessarily deriving from traditional PS */s/), except in some environments where the sibilant was in contact with a high vowel or a reflex of one of the corresponding semivowels. In these environments the writing of Egyptian <s> was much more heavily favored over <s> than in other environments.

For the present study a somewhat more rigorous approach has been adopted. The following reviews and review articles have been scrutinized for critical assessments of Hoch's (1994) interpretations: Gordon 1995, Quack 1996, Rendsburg 1996, Schneider 1996, Ward 1996, Kitchen 1997, Meeks 1997, Vittmann 1997, and Rainey 1998. The combined efforts of these scholars cast doubt on nearly half the relevant examples in Hoch's corpus. If all this material is simply discarded, the conclusion reached in the pilot study is still supported, but by a drastically reduced material base. On the other hand, simply accepting all the criticisms of Hoch's interpretations is not a viable option, because a few scholars seem to want to veto all interpretations based on the less frequent Egyptian perception of Semitic /s/ as /s/, a procedure that would instantly deprive the observation being presented here of its entire material base. Nor is such a procedure justifiable, since it flies in the face both of the evidence of the shibboleth story itself and of the scholarly consensus (which is supported by the evidence) that Semitic material came to the notice of the Egyptians not from a single uniform dialect but from a multiplicity of sources. (22) Consequently, the relevant criticisms and some other matters will be reviewed in detail (3.2. 1f.) before presenting the more compact data on which the hypothesis now rests (3.2.3f., 3.2.6).

Before proceeding to the detailed analysis of these criticisms, some general comments are in order.

The focus throughout has been on whether a given item can reliably be regarded as reflecting Egyptian perceptions of a particular area of Semitic phonetics. This has affected the approach that has been adopted with respect to group writing of apparently native lexemes and alternative etymologies.

Hoch has been criticized, sometimes justly (e.g., by Meeks 1997: 44 no. 246), for over-reliance on group writing as an index of foreign origin. Yet the use of many biliteral signs and groups to represent only their first consonant, which eventually became standard practice in Demotic (Vittman 1996: 446), seems not to have been standard in Late Egyptian. Consequently the view adopted here as a working principle is that while an occasional lapse into group writing of a native item proves nothing, more or less constant use of group writing for a particular lexeme indicates that the word was felt to be somehow out of place, whether due to obsolescence, neologism, real or perceived foreignness, etc. If, for instance, in a case of apparent obsolescence or reborrowing (23) the word has in addition a sufficiently good Semitic etymology, it has been assumed that association with a known Semitic vocable is responsible for the group writing and that the item can therefore be safely used for the purposes of this study. Any item that does not have such an etymology has been excluded, as have also hapaxes that fall into this category and items that seem not to be written "syllabically" at all.

A good competing Semitic etymology that casts doubt on the identity of the source sibilant or a good non-Semitic etymology casting doubt on the provenience of the item has generally been sufficient to eliminate the item in question from consideration.

Finally, items with initial <s> have been excluded, if it was felt that their causative status was sufficiently clear to have resulted in replacement of the Semitic causative marker /s/ by the Egyptian causative marker /s/.

3.2.1. The items in Hoch (1994: 17-396) to be reviewed are arranged below in order of Hoch's item numbers. The item number is followed [in square brackets] by Hoch's 5-high level of certainty judgment concerning the Semitic source; (24) then come one or more of Hoch's romanizations of the Egyptian, followed by Hoch's Semitic interpretation(s) introduced by an asterisk, and then an indication of Hoch's gloss (PN = occurs in name of person or god). This is followed by the decision: to include or exclude (the item from the study) plus brief discussion and indication of author(s) of relevant view(s), etc. The reviews and review articles listed above (3.2) have been referred to for the most part solely by author's name when the items they discuss are identified by Hoch's item numbers and arranged accordingly. Here follows the list.

No. 27 [2]: 'as=wa=ta. *'aswata; 'aswita. long plank.--Exclude: may not be group writing (cf. Meeks).

No. 35 [2]: 'i-sa-f. *'isafa? 'itafa? to scorch (a town); searing (of blows).--Include (pace Rainey): the item is at least Nineteenth Dynasty (post-1300) and thus late enough to reflect merger of PS */t/ with */s/ which is recorded in the El-Amarna letters and northern Canaanite dialects by ca. 1400 B.C. (Faber 1992: 3).

No. 72 [5]: '=p=s[i.sub.2]=ya(=t). *hippusita? beetle; grasshopper.--Exclude: sibilant and source uncertain. Judging by Hoch's clear (and sensible) preference for deriving the item from the root hps (and not from hps), the s in Hoch's Semitic reconstruction is a misprint. This item could then represent the solitary level [5] example of transcription of Semitic /s/ by Egyptian <s> alluded to in Hoch's (1994: 433, 436) tables, though it must be said that elsewhere Hoch denies this, claiming that no. 212 [5] "would be the only certain example" (Hoch 1994: 165 n. 181) of transcription of Semitic /s/ by Egyptian <s> if a (quite possible) derivation from Semitic mst "comb, card" were to be accepted. The evidence in some MSS for replacement of this apparently unfamiliar form with a familiar one (*p[??]-s[??]i(t) "the pig") may signal obsolescence (cf. Meeks), suggesting agreement with Ward's suggestion that the Semitic words may have been borrowed from Egyptian; alternatively it may point to a non-acclimatized item and thus a reborrowing, but if so the question remains: of what?

No. 102 [3]: 'a-g[u.sub.3]-su. *'agusu (sic: see discussion). belt.--Include. The Semitic reconstruction given is presumably computer error for *'aqusu, given the cited Hebrew cognate 'qs (the usually cited Arabic cognate reflects /s/ [see Hoch's 1994: 83 n. 142 cogent explanation of this], the Syriac /s/, which would be represented in Egyptian by <d> and <t>, respectively; Hoch 1994: 432f.) and the fact that plain s and acute-accented vowels have no place in Hoch's Semitic reconstructions, aside from a few clear errors like this one (and see next item, no. 166). Meeks's slight attack on the semantics does no harm to Hoch's etymology. Although the word occurs in Middle Egyptian (Quack), Hoch lists seven occurrences of full group writing from four different sources versus only one case where nothing but the final syllable is group written, all of which, together with a tolerable Semitic etymology, suggests that the word is usable as an index of Egyptian perceptions of Semitic phonetics.

No. 166 [2]: man=nu=sa. *manusa (sic: manusa; see discussion of preceding item no. 102). type of scribe.--Exclude. Quack's interpretation of the word (as meaning "Cretan") explains the group writing and is as good as anything offered by Hoch.

No. 198 [5]: ma-sa-hi. *masiha. amphora, a large vessel for wine and sesame oil.--Exclude, together with next, because of uncertain vocalism. Apart from s = s, Rainey objects to the displaced /i/ and the Neo-Babylonian date (i.e., post-1000 B.C.) of the Akkadian equivalent of this Twentieth Dynasty (i.e., ca. eleventh century B.C.) form, which could indicate a possibly tolerable gap of around two centuries, except that according to von Soden (1965-1981 s.v.), it is Late Babylonian (i.e., post-625 B.C.), which raises more serious doubts about the vocalism of Hoch's reconstruction, especially as it does not quite tally with the Egyptian writing. On the other hand the semantics of the equation are impeccable, even to the cubic capacity. Possibly the Egyptian represents an otherwise unattested construct state *masahhi "measuring vessel; measure of" (root msh), if a morpho-semantic parallel can be seen in eleppum "ship" beside elepu "to sprout" (cf. the ship's mast "sprouting" from the sea) and/or in agammum "marsh" beside agamu "to rage" (perhaps due to fiery reflections of sun on the surface of the water, cf. the etymology of some Indo-European words for wetlands, e.g., Old Church Slavonic blato "lake," Russian boloto "marsh," Lithuanian baltas, balas "white," Greek phalos "white," Sanskrit bhalam "splendour, lustre," Old Irish oibell "glow, heat," etc., Pokorny 1959: 118f.).

No. 199 [5]: ma-sa-hi-ta. *masihata? amphora.--See preceding.

No. 202 [4]: mas=t[i.sub.3]=ra. *mastira.--Exclude. Quack and Rainey present equally good Semitic etymologies based on the roots str and sdr, respectively, which fit the Egyptian <s> just as well.

No. 203 [5]: m[a.sub.4]=sa=ta=ha. *mashita. trap, snare.--Include. An ingenious piece of reconstruction by Hoch despite (somewhat emotional) comments by Rainey.

No. 207 [3]: ma=s=r=r[u.sub.2]. *Mesaru. plain; wetland.--Exclude: derivation of isrw, the lake of the lioness goddess (Meeks); and the word is not obviously in group writing.

No. 212 [5]: ma=sa=di=di=t. *mustata? comb.--Exclude: despite evidence that the word may be a reborrowing from Semitic (it is attested in Middle Egyptian--Quack; Meeks; there are two distinct instances of full group writing from different sources; and there also appears to have been a noteworthy change of meaning--Meeks), Arabic masata "comb (hair)," Akkadian masad/tu "comb hair; card wool" suggest that the Semitic sibilant in question was /s/, not /s/, despite Hoch's somewhat contradictory misgivings about this (cf. the discussion of no. 72 above). (25)

No. 229 [3]: ma=ga=sa. *maqasa. a metalworking activity: to emboss, hammer(?).--Exclude. May be three separate words, in which case group writing may be due to inadvertence; whether the P.Turin PR 32, 7 form means "mks-sceptre" (Meeks) or "document case" does not matter much since both words occur in Middle Egyptian.

No. 246 [3]: n[a.sub.2]+'a=sa. *na'asa. fierce, raging.--Exclude together with next. The word is known from at least Middle Egyptian (Meeks), and even though group writing is recorded from two separate locations (albeit mostly at Medinet Habu), the argument for a Semitic etymology is not strong.

No. 247 [3]: n='=s='u. *na'asa? to overpower.--See preceding.

No. 257 [2]: na=ha=sa. *nahasa. medicine for coughs.--Exclude. Ward is probably right that it is a miswriting of nhd.t nt s[??]y "id."

No. 274 [5]: ra=bi=sa=ya. *labisa. cuirass; leather armour.--Include as one with next; insufficient evidence for separation into two separate items (Quack).

No. 275 [5]: r[u.sub.2]=bi=sa. *labisa? to wear a cuirass.--See preceding.

No. 366 [2]: sa=b[a.sub.2]=-r. *samra? tapla? strong-smelling beer (or something contained therein).--Exclude: uncertain etymology not helped by Kitchen's cogent semantic suggestion "aroma."

No. 369 [3]: s[u.sub.3]=m[i.sub.3]=n. *sumelu // or *samna. north; left // or oil.--Exclude: uncertainty of etymology compounded by Rainey's convincing restoration of Semitic tmn "eight" as a possibility.

No. 372 [4]: sa=r[u.sub.2]=na. *sarona. plain.--Exclude: initial may represent */t/ (Rainey; root correctly specified in no. 207).

No. 373 [3]: s=-r=hut=ta. *saluhata. stalks, bunches.--Include. Rainey's alternative etymology (Arabic sariha "slice") is semantically incompatible with the context, even if one tries to extrapolate "slice" [right arrow] "cut" [right arrow] "reap" [right arrow] "harvest."

No. 374 [4]: sa=-r=hi. *salliha? to strip off.--Exclude: good alternative etymology in Arabic saraha "crack, splinter" (Rainey).

No. 377 [3]: s=ha=q=q. *sahaqiqu? PN the growler (?).--Exclude: possibly some Semitic influence, but alternative etymology (Quack) makes native Egyptian causative marker /s/ probable.

No. 378 [2]: s=hi=na=sa. *sahinasa. to be stirred up; provoked.--Treat the two sibilants separately. Include for medial sibilant. Exclude for initial: causative marker may be native Egyptian.

No. 379 [3]: sa=ha=ra='a. *sahli'a? to scorn; hold in contempt.--Exclude: despite good case for Semitic influence this appears to be a native word with native causative marker (Quack; Ward).

No. 387 [4]: sa='=ra. *ta'ara. calculation; scheme; or threat, promise.--See no. 389 (cf. also Meeks and Quack for divergent semantics).

No. 388 [3]: sa='a=ra. *ta'ara. to scheme.--See next.

No. 389 [5]: sa='a=ra. *sa'ara. market price.--Include as one together with nos. 387, 388 as different manifestations of the same vocable (Rainey; cf. Meeks for semantics).

No. 397 [5]: s[i.sub.2]=b=d[a.sub.2]. *sibta. staff; rod.--Include. Quack's refinement of the semantics (axe- or knife-handle) simply replaces one long, narrow piece of shaped wood with another.

No. 406 [5]: sa=-r=ma. *salama? sallema? to greet; make obeisance; do homage.--See no. 408.

No. 407 [5]: sa=ra=ma. *salama? sallema? to lay down (arms); seek peace.--See next.

No. 408 [5]: sa=ra=m[a.sub.4]. *salama? peace; greetings.--Include as one with nos. 406, 407.

No. 418 [4]: sa=da. *sadda. to assail, assault, beset.--Exclude. An isolated group writing; two possibilities for identification with native Egyptian words have been suggested (Quack; Ward).

No. 453 [3]: ka=wi=sa=na. *kusana? chariot saddle pads or reins?--Exclude. As in the cases of nos. 72 and 212, Semitic sibilant may be either /s/ or s/ (not mentioned by Hoch's reviewers).

No. 509 [5]: ga=wa=sa. *qawasa. to be crooked; to turn away.--Include as one with next. Even though a word of related meaning occurs in Middle Egyptian (viz. gws "loucher"--Meeks), verb and noun are found group written in seven different sources (versus one occurrence not in group writing) and the Semitic etymology is very sound both phonologically and semantically.

No. 510 [5]: ga=wa=sa. *qawasa. crookedness.--See preceding.

No. 522 [4]: gas=mu. *gasma. tempest; rainstorm.--Exclude. Solitary group writing probably of little significance: the word h[??]nw "waves" in Hoch's context must be group writing for Middle Egyptian hnw "id." The determinatives suggest the word means a body of water, not a storm (Rainey). Even though Hoch indicates that the connection with Demotic gsm "tempest; anger" was accepted by a number of earlier scholars--also accepted more recently by Vittmann (1996: 445)--and was followed by Erichsen (1954: 593), the Demotic word is clearly written with the wind determinative t[??]w (Erichsen 1954: 669). More comparable with our Late Egyptian word is Demotic gsm[??] "land irrigated by canals" with the same body-of-water determinative and apparently a compound of Demotic gs "side; *district" (for the semantics cf. Russian storona "side; district; parts") and m[??] "canal," which suggests that Hoch's gas=mu is really gs-mw "wetland." The resulting interpretation of the context, something like "The wetland produces waves," also appears acceptable, unless of course the scribe wrote the wrong determinatives under the influence of a word such as this.

No. 535 [3]: ta=ha=sa. *dahasa/da'asa. to crush, pulverize.--Exclude. Semitic sibilant very uncertain; the word may go back to Old Egyptian ths "stretch, soften" (Meeks).

No. 537 [5]: t[a.sub.3]=h=ba=sa=ta. *tahbasata. a type of basket.--Exclude: probably Hurrian (Meeks; Quack).

The above reduces Hoch's numbers of cases reliably attesting Semitic /s/ from forty-nine to thirty-eight written with Egyptian <s> and from thirty-five to nineteen written with Egyptian <s>. (26) Further items which have been included in the study even though they may appear to be a source of concern are as follows.

No. 28 [3]: 'as=ba=-r. *'aspara? whip? Hoch points out that this and no. 153 ([3] pa=-r=sa. *parata? split open; tear) come from a pair of roots prs and prt. Rainey comments that it would be more sensible to derive no. 153 from prs as well, but in the absence of further information it has been deemed advisable to retain the decision in this pair reached by Hoch.

No. 402 [5]: sa=ma=sa. *samsa. the sun, the sun god Samas. Old South Arabic and Arabic point to the original etymological initial being PS */s/, but the Hebrew and Aramaic forms show (and other forms cited by Hoch suggest) that assimilation of this initial to the stem-final /s/ was widespread in Semitic (cf. also Faber 1980: 176; 1981: 238).

Like no. 35 (3.2.1 above), nos. 206, 401, and 403 appear to have had etymological /t/, but attestation not earlier than the Twentieth Dynasty makes them suitable, as Hoch himself indicates, for inclusion as items representing living Semitic /s/ (details in 3.2.6).

3.2.2. Of the thirty-eight plus nineteen items of 3.2.1 (above), three (one plus two) cannot be used to confirm or deny our thesis regarding vocalic environment because their vocalism is too uncertain. They are:

No. 93 [5]: 'a=sa=qu. *'asaqu? 'asuqu? acts of oppression.

No. 365 [4]: sa=b[a.sub.2]=-r. *sabala? subla? spike, cluster.

No. 368 [4]: s[u.sub.2]=m[a.sub.2]='i(?)=n. *Sim'on. PN hearing; with metathesis of the vowels "*som'in for [sim'on]."--The spelling could also represent a prototype *sum'an (Woodhouse 2002: 178-80) or a metathesized *som'an- < Canaanite *sam'on- < PS *sam'an-, which is Kutscher's reconstruction cited with approval by Koehler/Baumgartner/Stamm (1995: 1457). The analysis below indicates preferred reconstructions for two of the above, viz. *sabala for no. 365 and *sam'on-/*som'an- for no. 368.

3.2.3. The remaining thirty-seven <s>-items plus seventeen <s>-items supporting our thesis that Semitic /s/ in high vowel (/i, u/) environments is denoted overwhelmingly by Egyptian <s> are classified in the tabulation in 3.2.6 below, according to the Egyptian sibilant and its phonological environment. Note that "high vowel environments" include cases where the sibilant is immediately preceded by a vowel originating in a diphthong having /y/ or /w/ as its second element (naturally a vowel of this type does not count as high vowel environment for any sibilant it comes after). In stem-final environments, only vowels preceding the sibilant have been taken into account since vowels can vary with inflectional endings, disappear in status constructus, etc.

It may also be noted that where the vowel follows Egyptian <s>, its specification is entirely a matter of editorial interpretation since in the relevant material, with one exception (no. 413, which allegedly indicates no vowel after the sibilant), the same sign group is written (variously transcribed sa, s[i.sub.2] and s[u.sub.5]), whatever the vowel.

3.2.4. Scrutiny of sections A.A.A, A.B.A, B.A.A and B.B.A of 3.2.6 below reveals, for high vowel environments in the word-initial and stem-medial positions, a clear preponderance--in the ratio 9:1--of Egyptian writings with <s> (27) (five and four, respectively) over those with <s> (nil and one, respectively). This may be compared with ratio 2:1 in the corresponding non-high vowel environments covered by sections A.A.B, A.B.B, B.A.B and B.B.B (sixteen plus six against four plus seven, or twenty-two against eleven). These results are summarized in Table 1.

These figures, I submit, represent strong evidence in favor of our thesis regarding Egyptian perceptions of Semitic /s/ in non-stem-final positions. It may even be possible to explain away the solitary exception with written <s> followed by short high vowel, namely no. 53 [3] ya=m[a.sub.4]=s[u.sub.3]=ra. *Yamsula, as being due to the preceding consonant, cf. no. 61 [3] ya=k=sa=m[u.sub.2]. *Yaqsamu, or to the shortness of the following vowel.

The behavior in the stem-final position is clearly aberrant. The fact that "only" two of the five examples with <s> in section B.C have a preceding high vowel, as against three with non-high vowels, can hardly be counted as real support for our thesis in this position. The aberrant behavior is no doubt due to some kind of low level auslaut affect (such as lenition), which presumably countered to a considerable extent the high vowel effect that was elsewhere audible to the Egyptians.

3.2.5. Further (and unanimous) support for the conclusion just reached is provided by the nine items clearly attesting the similarly but even more definitely polarized Egyptian perceptions of Semitic /s/. As is clear from the discussions of 3.2.1 above and the tabulation in 3.2.7 (section A) below, all three items written with Egyptian <s> that were excluded from the main part of the study on the basis that their Semitic sibilant might be /s/ rather than/s/ are reconstructed by Hoch with the non-stem-final sibilant in a high vowel environment. This contrasts with the six items, all (with one exception that can be excluded from consideration) written with Egyptian <s> and securely reflecting Semitic /s/ in a non-high vocalic environment. In five further items, including no. 535, which reflects either /s/ or /s/, the written <s> is stem-final. The three items excluded here, with reasons, are:

No. 85 [3]: 'a=ra=sa=na. *harasana. a skin condition: scabies? itch?--Sibilant probably affected by no. 84 (see 3.2.6 A.B.A), which in some languages is related to similar words denoting a similar skin condition.

No. 88 [4]: '=s=ba. *'(i?)sba. grassy patch?--Uncertain vocalic environment (indeed the prototype could just as easily be *'asab-, corresponding to Hoch's Hebrew 'eseb "green plants").

No. 369: (see 3.2.1 above).--Both sibilant (/s/, /s/ or /t/) and vocalism (/u/ or /a/) are too uncertain.

3.2.6. The tabulation of Egyptian transcriptions of Semitic /s/. (28)
 A. Written with Egyptian <s>
 A.A. Word-initial sibilant
 A.A.A. Adjacent high vowel

b 395 [3] s[u.sub.5]=ba=ya. *subaya? a precious stone: agate?
bl 396 [5] s[i.sub.2]=b[a.sub.2]=r[u.sub.2]=t[a.sub.5]. *sibbolet.
 flowing stream, torrent.
b 397 [5] s[i.sub.2]=b=d[a.sub.2]. *sibta. staff; rod.
m 399 [4] s[u.sub.5]=ma. *suma. name.
bl 403 [3] s[u.sub.5]=n=b='u. *sulbi? trumpet; (kohl) tube.

 A.A.B. Adjacent non-high vowel
r 389 [5] sa='a=-r. *sa'ar(a). market price.
b 392 [3] s[u.sub.5]=w[i.sub.2]=b=ti. *so'ibta? a vessel.
w 394 [3] sa=wa=sa=ta. *sawasata? administration?
m 400 [5] sa=m='. *sama'a. PN to hear.
mr 401 [4] sa=m[a.sub.4]=ri=ya. *samriya. PN protection.
m 402 [5] sa=m=sa. *samsa. the sun, the god Samas.
fl 404 [3] sa=-n=ra=fi. *salapa. to be disheveled.
ml 408 [5] sa=ra=m[a.sub.4]. *salama. peace; greetings.
ml 409 [5] sa=-r=ma=ta. *salmata? delivery or provisions.
r 410 [3] sa=ra=ga=hi. *saharaqi? passions.
- 411 [5] sa=ha=qa. *sahaqa. dust cloud; meal; bread.
- 412 [3] sa=qa. *saqa? unknown (be impelled, urged?).
br 413 [3] s=ka=ba ra=ka=ba. *sakba-rakba. pair of millstones.
- 414 [3] sa=ku=na. *saqona/*sakuna? watering place.
r 415 [4] sa=ga=-r. *sagara. body of water (ditch, dyke?).
l 419 [2] sa=di=r[u.sub.2]=ta={t}. *sadilota? ravine, chasm.

 A.B. Stem-medial sibilant
 A.B.A. Adjacent high vowel
b 5 [4] '[a.sub.2]=bi=s=tu. *'abistu. hardtack, biscuit(?).
f 35 [2] 'i=sa=f. *'isafa? 'itafa? to scorch; searing.
' 84 [5] 'a=r=s[i.sub.2]=na. *'adasina. lentils.
mr 206 [3] m[a.sub.2]=sa=-r. *mesara? < pat(t)ura? furniture: table?

 A.B.B. Adjacent non-high vowel
' 92 [5] 'a=sa=q. *'asaqa. to extort, defraud; oppress.
bm 205 [5] ma=sa='aba. *mas'aba. watering place.
m 208 [4] ma=sa=ra=ra. *masarrira. to attach; to affix.
bm 209 [3] ma=sa=ka=ba(=yu). *maskaba? (sing.), maskabayu/a? (pl.). a
 state official (tax? customs?).
m 210 [4] m[a.sub.4]=sa=ka=ta. *maskatta. dwelling place.
w 394 [3] sa=wa=sa=ta. *sawasata? administration?

 A.C. Stem-final sibilant
 A.C.A. Preceded by adjacent high vowel
bl 274 [5] ra=bi=sa=ya. *labisa. cuirass; leather armour.
r 285 [5] r[u.sub.2]=sa. *rosa. peak, summit; head.

 A.C.B. Not preceded by adjacent high vowel
- 378 [2] s=hi=na=sa. *sahinasa. to be provoked.
m 402 [5] sa=m=sa. *samsa. the sun, the sun god Samas
- 441 [5] qid=s[u.sub.4]. *qidsu. sanctuary.
w 504 [5] ga=wa=sa. *qawasa. to be crooked; to turn away.

 B. Written with Egyptian <s>
 B.A. Word-initial sibilant
 B.A.A. Adjacent high vowel: nil

 B.A.B. Adjacent non-high vowel
wl 361 [4] s=wa=-r. *sawala? trapper [chariot horse's skirt].
m 370 [2] sa=mak=ta. *samakta? beam; girder.
l 373 [3] sa=r=hut=ta. *saluhata. stalks, bunches.
p 384 [2] sa=ga=b[a.sub.2]=ya=na. *sapakayna? body of water.

 B.B. Stem-medial sibilant
 B.B.A. Adjacent high vowel
ml 53 [3] ya=m[a.sub.4]=s[u.sub.3]=ra. *Yamsula. PN to rule; reign.

 B.B.B. Adjacent non-high vowel
pr 28 [3] 'as=ba=-r. *'aspara? whip?
b 59 [5] ya=sa=ba=h[a.sub.2]. *Yasbaha. PN to praise.
m 61 [3] ya=k=sa=m[u.sub.2]. *Yaqsamu. PN fowler (?).
m 201 [4] mas=ta(='[u.sub.3]=t). *masota. small oared galley.
m 203 [5] m[a.sub.4]=sa=ta=ha. *mashita. trap, snare.
- 259 [4] n[a.sub.2]=sa=k. *nasaqa. to put in proper array.
- 302 [4] ha=s=t=ka=ta. *hastaggata. to swerve; careen; rock.

 B.C. Stem-final sibilant
 B.C.A. Preceded by high vowel
' 102 [3] 'a=g[u.sub.3]=su. *'agusu (sic: *'aqusu?). belt.
r 286 [4] r=k=s[u.sub.2]. *rakusu. equipment; gear.

 B.C.B. Preceded by non-high vowel
' 101 [2] '=ga=sa. *'agasa (sic: 'aqasa? if root is same as in no.
 102). a baked good.
r 282 [3] ra=ha=s[u.sub.2]. *rahasu. bread, cake.
l 482 [4] ka=ra=s[u.sub.2]. *galasu. to frisk; leap.

3.2.7. The tabulation of Egyptian transcriptions definitely or possibly representing Semitic /s/ (see n. 28 above regarding the letters in the left-hand column).
 A. Written with Egyptian <s>
 A.A. Non-stem-final sibilant
 A.A.A. Adjacent high vowel

p 72 [5] '=p=s[i.sub.2]=ya(=t). *hippusita? beetle; grasshopper.
m 212 [5] ma=sa=di=di=t. *mustata? comb (for <s> in contact with
 following consonant see n. 27 above).
- 453 [3] ka=wi=sa=na. *kusana? chariot saddle pads or reins?

 A.A.B. Adjacent non-high vowel: nil
 A.B. Stem-final sibilant: nil
 B. Written with Egyptian <s>
 B.A. Non-stem-final sibilant
 B.A.A. Adjacent high vowel: nil
 B.A.B. Adjacent non-high vowel

'r 358 [4] sa='a=r[u.sub.2]/-r. *sa'aru/a. barley (field) or scrub
'r 359 [5] sa='a=ra=ta. *sa'rata. wool.
b 380 [4] sa=qa=ba. *sagaba. in PN to be exalted.
- 381 [3] sa=ka='u. *saki'u? a military officer, scout? guard?
- 382 [3] sa=ga. *saqqa. to open the way, break the trail.
- 383 [4] sa=ga. *saqqa. inferior type of cloth, garment or blanket,
 probably sackcloth.

 B.B. Stem-final sibilant
mrw 164 [2] ma=-n=-r=wa=s[u.sub.2]. *marwasu? (sic: marwasu? cf.
 misprinted plain s in nos. 102, 166 in 3.2.1 above). nest.
m 465 [4] ku=m[a.sub.4]=sa. *kumasa? cowardice, or sim.
r 483 [4] ku=ra=sa. *kursa? sack.
- 512 [4] ga=ni=sa. *nagisa? violence, injustice.
- 535 [3] ta=ha=sa. *dahasa / da'asa. crush, pulverize.

3.3. The above analysis leads to the conclusion that in the Semitic dialects with which the Egyptians were most familiar, the auditory effect of the sin phoneme was most conspicuously s-like in certain environments involving high vowels. Acoustic/auditory justification for this in the case of the lip-rounding normally associated with /u/ can be found in Faber (1986: 169); the palatalizing effects of the high front vowel /i/ are well known. The two effects can be compared with the well-known ruki-rule of satem Indo-European (Woodhouse 2002: 177 n. 4).

If, as seems likely, these Semitic dialects were Cisjordanian rather than Transjordanian, and the Transjordanian sin was more s-like and did not produce a conspicuously s-like auditory effect even in high vowel environments, then in these high vowel environments the difference between the two types of sin would be particularly exaggerated and thus lead to a greater likelihood that precisely in these high vowel environments--in the manner suggested by Hendel--Cisjordanians would interpret the Transjordanian sin as samekh /s/. This explains the recorded Cisjordanian interpretations of Transjordanian /s/ as samekh /s/ in the high vowel environments of sibbolet (or *subbultu, etc.) and *ba'alisa but as sin /s/ in the non-high vowel environments of nahas and sobi, where sin on both sides of the Jordan was still somewhat more s-like, as the wavering Egyptian transcriptions indicate, at least for the Cisjordanian side.


4.1. Our analysis makes it probable that Faber's shift PS *[s] > Canaanite /s/ had its beginnings in Cisjordanian high vowel environments (29) (rather than in the general vicinity of certain consonants, see n. 28 above) and that during the early part of the Late Egyptian period, before the time of the shibboleth incident, it reached an intermediate stage which persisted until at least as late as the reign of the Ammonite king Ba'alisa. (30)

4.2. Two of the Egyptian transcriptions are of special interest in the present context. One is Hoch's no. 396 [5] s[i.sub.2]=b[a.sub.2]=r[u.sub.2]=t[a.sub.5], which Hoch interprets as *sibbolet "flowing stream; torrent," though it could equally well be s[u.sub.5]=b[a.sub.2]=r[u.sub.2]=t[a.sub.2] representing the prototype of Talmudic Aramaic subbalta' "current," which Hoch also cites.

The other is no. 365 [4] sa=b[a.sub.4]=-r. *sabala? *subla?, corresponding to Arabic sabl, Hebrew sibbolet, Talmudic Aramaic subbalta', etc., "ear of grain." As noted above (, our analysis indicates a preference for the reconstruction *sabala.

We can now appreciate that the phonological structure of the Gileadite test word was particularly appropriate for the purpose of the test since, as the Gileadites must have known, it presented the problematic sibilant in an environment of maximum contrast between their own and the Ephraimite realization of it. The actual meaning of the test word may not be important except that if the fugitive Ephraimites were forced to forage for food, then "ear of grain" might be just as appropriate to their desperate situation as "stream." (31) There seems no reason why it cannot represent both and thus stand as a long-forgotten testament to the battle wit of the Gileadite soldiery or their leaders--unless it was simply a popular Transjordanian shibboleth of the period.
Table 1: Numbers of cases of Semitic /s/ written with Egyptian <s> and
<s> in word-initial and stem-medial environments

adjacent written <s> written <s> ratio
vowel initial medial total initial medial total <s>:<s>

 1 1
high 5 4 9 0 1 1 9:1
non-high 16 6 22 4 7 11 2:1

My heartfelt gratitude is here expressed to Professor Alice Faber for many constructive criticisms of an earlier attempt of mine to deal with this problem. The usual disclaimers apply.

1. According to the source, 42,000 Ephraimite (i.e., Cisjordanian) fugitives were slain one-by-one upon mis-pronouncing as sibbolet the word sibbolet (which may have meant "ear or grain," "olive branch," "stream," "torrent," or "flood") put to them as a test by their Gileadite (i.e., Transjordanian) captors.

2. On Hendel's (1996: 70) relegating Faber to the differential phonemic merger school see below, 2.4.

3. For a generally satisfactory overview of earlier research see Emerton 1985. Emerton's (1985: 155f.) principled revival of the possibility of mere phonetic difference as the explanation appears to have been significant in the development of Hendel's view.

4. On Rendsburg's wishful phonology and etymology. see Faber 1992: 2f., Hendel 1996: 70.

5. See Faber 1980: 171-227; 1981; 1982: 83; 1984: 190-92 (note that on p. 190 "[S.sub.1] = /hl/" is a misprint for "[S.sub.1] = /s/, [S.sub.2] = /hl/"); 1985: 105; 1986: 164f.

6. E.g., in the present context it might have compounded the joke rather nicely if the starving, hunted Ephraimites, on being asked to say "ear of grain" or "river" (in full view of the Jordan), had been heard to respond with something that sounded a bit like sib'a "plenty" or saba't "I'm full," or sepuna "it is hidden."

7. Though on rereading, even these plurals become suspect because two pages later (Faber 1992: 4 line 2) the reader is confronted with the disconcerting misprint "Gileadites" for "Ephraimites."

8. Cf. Faber 1980: 171-80; 1981: 233-41 on the loss of Semitic /s/ beside--for the record--Faber 1986: 164-67 on the emergence of */s/ in pre-Semitic.

9. See Hoch 1994: 368, where no. 548 seems to be the latest example of this (but see also n. 10 below).

10. It may be objected that the writing of samekh as an affricate in Late Egyptian transcriptions down to at least the eighth century means that only phonological systems with affricate samekh should be considered in the present case. Actually there is no problem with doing this at all (see n. 13 below). On the other hand it is also possible to argue against so late a date for the final deaffrication of samekh, as follows. First, the Egyptian transcriptions collected by Hoch (1994: 17-396) tend to show a great conservatism in their phonemic representation of any given lexeme. For example, of some sixty-seven entries in which Semitic /t/ and /s/ are transcribed sometimes with Egyptian <s> and sometimes with <s>, and all of which show some orthographic variation, only two entries show variation in the actual choice of sibilant. Consequently, first attestations can generally be regarded as fixing the Egyptian phonetic perception of the Semitic word to be used in subsequent centuries. Secondly, for all but one of the lexemes involving samekh, the first transcription is attested no later than the Twentieth Dynasty, i.e., no later than ca. 1100 B.C., and the one exception (Hoch's no. 548), since its contact with the group writing system is distinctly minimal, can be regarded as an old borrowing taken into Egyptian well before this cutoff date. Perhaps this date for West Semitic deaffrication of samekh can be pushed back still further: Muchiki (1999: 277) cites two Ugaritic transcriptions in which Egyptian /s/ is represented by samekh, although as we have seen this is not necessarily a guarantee of deaffrication. (Apart from the El-Amarna transcriptions, which show typical Old Babylonian features [Huehnergard 1998: 63; for Old Babylonian, cf. Sommerfeld in von Soden 1995: 36], the rest of Muchiki's material appears to be post-1000 B.C.)

11. This affricate samekh is perhaps the chief mark of conservatism; it allows PS fricative sin to be in principle unspecified for any of the features we are currently considering, although Faber 1982: 83; 1986: 169 reconstructs PS sin specifically as apical and, we may infer, narrow-groove.

12. For example, if the progressive dialect B were Gileadite, the Ephraimites speaking dialect A might hear the /s/ of B as practically identical with their own /s/ by virtue of the possible "palatal quality" of apical /s/ (Stevenson 1970: 16; this may refer to a tendency for this sibilant to be slightly retroflexed or to be produced with a certain amount of wide-groove friction as well). But the Gileadites would recognize that the Ephraimite response did not contain a fully wide-groove articulation. This corresponds to Faber's interpretation in its first variant. Vice versa, if the Ephraimites speak dialect B and respond to the conservative Gileadite narrow-groove /s/ with their own narrow-groove /s/, because that is the sound their ears tell them they are hearing, the Gileadites will pick up the laminality of the response, yielding Hendel's interpretation.

13. It should be emphasized that the hypothetical pair proposed here is only one of several configurations of the given features that could lead to the same result and at the same time satisfy the perceived typological requirement that both dialects have a phoneme /s/. For example, if affricate samekh is insisted upon in both dialects--preferably narrow-groove laminal in both--both dialects can still have narrow-groove sin, apical in one, laminal in the other, and the choice of Gileadite versus Ephraimite assignment will still be completely free; but as usual one scenario will have the Ephraimites simply responding with their own sin, while the other will have them cannily responding with their nevertheless inadequate samekh.

14. This, incidentally, is not the same as asserting that speakers of both dialects would be equally familiar with the speech of the other (which is not necessary to the point being made here) or that the Ephraimites would be in a position to mimic successfully the Gileadite pronunciation, as one reviewer has claimed. After all, the whole point of Hendel's interpretation (and of one version of Faber's) is that both Gileadites and Ephraimites recognized that there was a difference in their pronunciations of the test word and that the Ephraimites were unable to mimic successfully the others' pronunciation. (No doubt the Gileadites would have been equally unable to mimic successfully the Ephraimite pronunciation, but the Gileadites were not the ones being put to the test.) All that is required is that speakers of both dialects are able to recognize that speakers of the other dialect speak "with an accent" and that they believe they can identify a particular difference and mimic it. This ability and this belief is surely the common property of humanity and is not confined to those with linguistic or psycholinguistic training.

Successful imitation of a speech difference is something quite different, signaling an ability possessed by relatively few and requiring perhaps a special gift or early childhood exposure or extensive coaching (and probably not helped much by ordinary linguistic training). Its rarity is such that is seems not to have been achieved by any Ephraimite in the story--certainly we are not told of any. Incidentally, the idea that the two groups of speakers in the story were in reasonably close linguistic contact seems to be borne out by the cause of their conflict, namely that the Ephraimites were incensed that the Gileadites had not called upon them to join them as allies in a joint mission against the Ammonites (Judges 12:1).

15. It is curious that Hendel 1996: 70 prefaces what can be read as a correct statement of Faber's position with three incorrect "merger" statements of it, all on the same page: once in a table summarizing previously proposed solutions where the incorrect summary "*[s.sub.1] [right arrow] s in Gileadite; *[s.sub.1] retained in Ephraimite (Faber 1992)" appears in a column headed "Phonological mergers"; and twice in the text, the precise misstatements being "Blau and Faber posit different phonological mergers involving *s in Ephraimite and Gileadite Hebrew" and, equally wrongly, "Against Blau and Faber, there is no evidence outside their theories of sibbolet: sibbolet for differing phonological histories of the phoneme s on the two sides of the Jordan. The phonemic distinction of s : s is preserved in all known Northwest Semitic dialects of the Iron Age." Let us be clear: the development of [S.sub.1] into /s/ is emphatically not a merger of [S.sub.1] with sin or samekh ([S.sub.3]), contrary to what Hendel's mistaken treatment may seem to suggest.

16. The other information presented in Rendsburg 1988a: 258 and in 1988b: 76 n. 15 concerns the writing of etymological /t/ as <s> occasionally in Amorite and consistently in Tell Fakhariyeh Aramaic. This information does indeed show or confirm that such a practice was possible in the Semitic world, but by itself it in no way compels the view that PS */t/ was retained intact in eleventh-century Gileadite or sixth-century Ammonite.

17. This tag should probably be changed to exclude South-West Semitic and include Babylonian.

18. The latter may be the more correct if Millard (1983) is right in perceiving the same root in mt''l, since this royal name is transcribed ma-ti-'-AN in Neo-Assyrian (Gibson 1975: 34), but this may simply reflect the Aramaic development /t/ > /t/.

19. For a critique of Rendsburg's treatment of Ammonite material both Hendel (1996: 72) and Faber (1992: 3) refer the reader to Beeston (1988), who also deals only with b'lys', but does so unsatisfactorily. For, seemingly oblivious to Rendsburg's (1988a) apparent lack of logic, Beeston attempts an alternative etymology of the name, an attempt that cannot be taken seriously because it relies on an apparent misreading (not a word is said about emendation!) of the final 'ayin of the name as 'aleph (the 'ayin is perfectly clear in Herr's [1985: 169] photograph).

20. The displacement of the sign SI by SI in Old Assyrian (Gelb 1974: 97f.) suggests that at least one of the processes that led to the crossover of /s/ and /s/ in Assyrian (Millard 1974: 4, 7) began in the vicinity of /i/.

21. Since this explanation concerns differing Cisjordanian perceptions of Transjordanian West Semitic /s/, there seems little point in including Assyrian material with its complete crossover of /s/ and /s/. Moreover, judging by Rendsburg's (1988b: 74) examples, the Assyrians heard all West Semitic /s/ as /s/ irrespective of vocalic environment, cf. -si- in me-na-si-i)= mns) beside sa- in sa-mi-ri-na (= smrn), etc. Exceptions, such as the place names Aramaic skn = Neo-Assyrian [.sup.uru]si-ka-ni (Tell Fakhariyeh 11. 13/20) and Aramaic qst = Neo-Assyrian [.sup.uru]qastu(BAN) (Postgate 1973: 35), appear to involve items belonging to both traditions, Sikan being near a possible meeting place of East and West Semitic culture (Greenfield/Shaffer 1983: 111), while qst / qastu signifies "archer's bow" in both languages.

22. See Schneider 1996: 176 for a detailed list of those new phonological correlations of Hoch's that Schneider regards as securely established.

23. Certainly outside Afro-Asiatic the reality of reborrowing and sequential borrowing, which Hoch 1994: 6 n. 16 acknowledges, is easily recognized in such cases as, e.g., French riche, Italian ricco, etc., "rich," which go back through Germanic and Celtic to the same stem as Latin reg- "king" (see, e.g., OED s.v. rich), the immediate source of French roi, Italian re "id.," and English gentle, genteel, and jaunty representing successive adoptions (in the thirteenth, fourteenth, and seventeenth centuries) of French gentil (OED s.vv.).

24. On the subject of using material from Hoch's lower levels, cf. Kitchen's (1996: 90) expression of "Happy satisfaction" with the overwhelming majority ("about 480 out of 500") of Hoch's Semitic identifications, whatever their level of certainty.

25. Hoch's (1994: 433, 436) tables of Semitic/Egyptian transcriptional correspondences indicate one level [5] and one level [4] transcription of Semitic /s/ by Egyptian <s>. By contrast, Hoch's alphabetic listing with transcriptions of the material reveals only one level [3] example of this, viz. no. 85 (see 3.2.5 below). As has been indicated above, both nos. 72 and 212 qualify as the level [5] entrant. One of these may have been the intended level [4] example, though no. 85 might also qualify for this (no. 453 [3]--below--is far too uncertain), since Hoch's discussion of the preceding item (no. 84 [5]) suggests an earlier view that nos. 84 and 85 were identical, a view replaced in the discussion of no. 85 by the suggestion that no. 84 may have influenced the sibilant of 85 (while Hoch's footnote 181 (p. 165) seems to give both views equal weight). Before this apparent change of heart, it is possible that no. 85 may have been assigned a higher level of certainty, i.e., level [4]. Otherwise all secure cases of Semitic /s/ are transcribed by Egyptian <s> (for details, see 3.2.5 and 3.2.7 listing B below).

26. Some of Hoch's reviewers also furnish additional data ostensibly of relevance for the present study, namely 'iksr/grs "type of snake" (Quack 1996: 512), mhbs "an object made of ivory," isknw "sashes(?)" and ssm(t) "horse" (Meeks 1997: 56). The first three of these items contain Egyptian s, so that whatever vocalism may be assigned to them they cannot conflict with the results of the present study. The fourth, ssm(t), is clearly to be compared with Akkadian sisum (Old Akkadian, Old Babylonian [.sup.anse]si-si. Old Assyrian sisium, sisa'um, Middle Babylonian also si-si-ka, etc.). Ugaritic ssw, Old Aramaic, Hebrew sus, all meaning "horse" and pointing generally to samekh. This, plus the fact that the origin of the item is unknown (von Soden 1965-1981 s.v.; cf. Koehler/Baumgartner 1958: 651 and Koehler/Baumgartner/Stamm 1983: 704f. s.v. sus; Mayrhofer 1956: 62 s.v. asvah, 1960: 140f. n. 35, and 1992: 136f. s.v. asva-), places it outside the scope of the present study.

27. That the writing of no. 5 [4] '[a.sub.2]=bi=s=tu. *'abistu with <s> is not due to the immediately following consonant is suggested by nos. 59 [5] ya-sa=ba=[a.sub.2]. *Yasbaha, 203 [5] m[a.sub.4]=sa=ta=ha. *mashita, 302 [4] ha=s=t=ka=ta. *hastaggata, etc.

28. The explanation of the left-hand marginal column in the tabulation is as follows. Since Faber (1986) has argued that the presence of labials, liquids and 'ayin can be responsible for a shift /s/ > /s/ in the same word--and indeed both sibbolet and b'lys' contain some of these consonants (two [b, l] and three [b, l, '], respectively)--the opportunity has been taken here to highlight the presence of these consonants in items in the tabulation by noting the relevant consonant(s) in the left-hand margin. The three items excluded in 3.2.2 also contain some of these critical consonants, thus no. 93 (with <s>) has '; no. 365 (with <s>) has b, r; no. 368 (with <s>) has m, '. As can be seen from these three and a glance at the tabulation, of the thirty-seven items written with <s>, all but five (in their order in the tabulation: nos. 411, 412, 414, 378, 441) (or six if no. 453 is included from 3.2.7) and of the nineteen with <s>, all but two (nos. 259, 302) contain at least one of these critical consonants. It is clear from these figures that the consonants in question have had no effect on Egyptian perceptions of Semitic /s/ (and certainly not in the sense predicted by Faber but, if anything, the opposite).

29. Cf. also fourteenth-century El-Amarna l[a-k]i-si ~ la-ki-si (Woodhouse 2002: 180f.).

30. Some confirmation of this late date may be detectable in the alphabetically arranged list of Semitic words written in Demotic and collected by Vittmann (1996), although the variety of source languages is a difficulty and may account for the fact that the reflexes of traditional PS */s/, and also of */s/, are transcribed mostly by Demotic <s>, although occasionally also by Demotic <s>. Steiner's (2000: 191) suggestion that the equation Semitic */s/ = Egyptian <s> may point to a Phoenician/Punic origin for the Tebtunis material is not of course binding on all of Vittmann's data. The one clear example of <s> representing /s/ seems to be Demotic gsm "storm, tempest," Coptic gosm (transliteration according to Walters 1972: 1) "darkness, tempest" : Hebrew gesem (< *gasam-) "heavy rain," Ugaritic gsm "rainstorm" (cf. Hoch 1994: 354 no. 522), which can be regarded as an independent Demotic borrowing if, as has been suggested above (3.2.1), it is to be separated from no. 522 gsmw "lake, wetland?" In that case Demotic gsm "storm, tempest" agrees with our hypothesis by presenting the sibilant in a non-high environment. This contrasts with the items written with Demotic <s>, for which both high and non-high vowel environments can be reconstructed, e.g.: wrsk "Ammoniaksalz," Coptic ousak: Arabic wussaq "gomme ammoniaque"; bsl "a saw," Coptic basour : Hebrew massor, Arabic minsar (cf. also Steiner 2000: 192); mnslg "splitting tool": Arabic salaqa "split," probably with high vowel prefix mi-; nsr "hawk, falcon," Coptic nosr: Hebrew neser (< *nasar-); mglst "mortar" : Arabic garasa (misprinted "garasa" in Vittmann 1996: 439) "grind, crush" (cf. also Steiner 2000: 192); and several others (including 'rsn "lentils" : Hebrew 'adasim, which, as we have seen (3.2.6 above), occurs already in Late Egyptian as Hoch's no. 84). Other examples with <s> may not reflect /s/ at all, such as Demotic s'rt "wool" with Semitic /s/, cf. Hebrew sa'arah, and Demotic kswr/gswr "ring," which appears to reflect Semitic /t/ (Hoch 1994: 355 no. 523). Even so, it appears we have in Demotic gsm "storm, tempest" a possible additional indication that Transjordanian /s/ continued to be heard by Cisjordanians as /s/ until at least the sixth century B.C.

31. Emerton (1985: 153) counters Swiggers's (1981) cogent data supporting the view that the "watery" meanings of the test word were appropriate to the situation with the example of the nonrelevant town name Scheveningen used by the Dutch to detect Germans during WWII. A possible difference here is that Gileadite and Ephraimite seem to have been mutually intelligible dialects, which is not, I think, the usual view of Dutch and German.


Abou-Assaf, Ali. 1981. "Die Statue des HDYS'Y, Konig von Guzana." Mitteilungen der deutschen Orient-Gesellschaft zu Berlin 113: 3-22.

Beeston, A. F. L. 1988. "Sibbolet: A Further Comment." Journal of Semitic Studies 33: 259-61.

Blau, Joshua. 1977. "'Weak' Phonetic Change and the Hebrew Sin." Hebrew Annual Review 1: 67-119.

Dobson, E. J. 1957. English Pronunciation 1500-1700. Oxford: Clarendon Press.

Emerton, John A. 1985. "Some Comments on the Shibboleth Incident (Judges XII 6)." In Melanges bibliques et orientaux en l'honneur de M. Mathias Delcor, ed. A. Caquot, S. Legasse, and M. Tardieu. Neukirchen-Vluyn: Neukirchener. Pp. 149-57.

Erichsen, W. 1954. Demotisches Glossar. Copenhagen: Ejnar Munksgaard [rpt. Milan: Cisalpino-Goliardica, 1972].

Faber, Alice. 1980. "Genetic Subgrouping of the Semitic Languages," Ph.D. dissertation, University of Texas.

________. 1981. "Phonetic Reconstruction." Glossa 15: 233-62.

________. 1982. "Early Medieval Hebrew Sibilants in the Rhineland, South Central and Eastern Europe." Hebrew Annual Review 6: 81-96.

________. 1984. "Semitic Sibilants in an Afroasiatic Context." Journal of Semitic Studies 29: 189-224.

________. 1985. "Akkadian Evidence for Proto-Semitic Affricates." Journal of Cuneiform Studies 37: 101-7.

________. 1986. "On the Actuation of Sound Change: A Semitic Case Study." Diachronica 3: 163-84.

________. 1992. "Second Harvest: Sibbol[??] Revisited (Yet Again)." Journal of Semitic Studies 37: 1-10.

Gelb, Ignace J. 1974. "An Old Assyrian Votive Vessel." In Anatolian Studies Presented to Hans Guterbock on the Occasion of his 65th Birthday, ed. Kurt Bittel, Philo H. J. Houwink ten Cate, and Erica Reiner. Istanbul: Nederlands Historisch-Archeologisch Instituut in het Nabije Oosten. Pp. 95-104.

Gibson, John L. C. 1975. Textbook of Syrian Semitic Inscriptions, 2: Aramaic Inscriptions Including Inscriptions in the Dialect of Zenjirli. Oxford: Clarendon Press.

Gordon, Cyrus H. 1995. Review of Hoch (1994). Catholic Biblical Quarterly 57: 770-71.

Greenfield, Jonas C., and Aaron Schaffer. 1983. "Notes on the Akkadian-Aramaic Bilingual Statue from Tell Fekherye." Iraq 45: 109-16.

Hendel, Ronald S. 1996. "Sibilants and Sibbolet (Judges 12:6)." Bulletin of the American Schools of Oriental Research 301: 69-75.

Herr, Larry G. 1985. "The Servant of Baalis." Biblical Archeologist 48: 169-72.

Hoch, James E. 1994. Semitic Words in Egyptian Texts of the New Kingdom and Third Intermediate Period. Princeton: Princeton Univ. Press.

Huehnergard, John. 1998. "A Grammar of Amarna Canaanite." Bulletin of the American Schools of Oriental Research 307: 59-77.

Kitchen, K. A. 1997. Review of Hoch (1994). Bulletin of the American Schools of Oriental Research 307: 89-91.

Koehler, Ludwig, and Walter Baumgartner. 1958. Lexicon in Veteris Testamenti libros. 2nd ed. Leiden: E. J. Brill.

________. 1983. Hebraisches und aramaisches Lexikon zum Alten Testament, 3. Lfg, 3rd ed., rev. Walter Baumgartner and Johann Jakob Stamm. Leiden: E. J. Brill.

________. 1995. Hebraisches und aramaisches Lexikon zum Alten Testament, 4. Lfg.

Marcus, Ralph. 1941. "The Hebrew Sibilant Sin and the Name Yisra'el." Journal of Biblical Literature 60: 141-50.

________. 1942. "The Word Sibboleth Again." Bulletin of the American Schools of Oriental Research 87: 39.

Mayrhofer, Manfred. 1956. Kurzgefasstes etymologisches Worterbuch des Altindischen, 1: A-TH. Heidelberg: Carl Winter.

________. 1960. "Indo-iranisches Sprachgut aus Alalah." Indo-Iranian Journal 4: 136-49.

________. 1992. Etymologisches Worterbuch des Altindoarischen, 1. Heidelberg: Carl Winter.

Meeks, Dimitri. 1997. "Les emprunts egyptiens aux langues semitiques durant le Nouvel Empire et la Troisieme Periode Intermediaire: Les aleas du comparatisme." Bibliotheca Orientalis 54: 32-62.

Millard, R. A. 1976. "Assyrian Royal Names in Biblical Hebrew." Journal of Semitic Studies 21: 1-14.

________. 1983. "Assyrians and Aramaeans." Iraq 45: 101-8.

Muchiki, Yoshiyuki. 1999. Egyptian Proper Names and Loanwords in North-West Semitic. Atlanta: Society of Biblical Literature.

OED 1971 = The Compact Edition of the Oxford English Dictionary. 1st ed. Oxford: Clarendon Press.

Pokorny, Julius. 1959. Indogermanisches etymologisches Worterbuch, 1. Munich: A. Francke.

Postgate, J. N. 1973. "Assyrian Texts and Fragments." Iraq 35: 13-36.

Quack, Joachim Friedrich. 1996. Review of Hoch (1994). Zeitschrift der Deutschen Morgenlandischen Gesellschaft 146: 507-14.

Rainey, Anson F. 1998. "Egyptian Evidence for Semitic Linguistics (Review of Hoch 1994)." Israel Oriental Studies 18: 431-53.

Rendsburg, Gary A. 1988a. "More on Hebrew Sibbolet." Journal of Semitic Studies 32: 255-58.

________. 1988b. "The Ammonite Phoneme /T/." Bulletin of the American Schools of Oriental Research 269: 73-79.

________. 1996. "Semitic Words in Egyptian Texts." JAOS 116: 508-11.

________. 1997. "Ancient Hebrew Phonology." In Phonologies of Asia and Africa (including the Caucasus), vol. 1, ed. Alan S. Kaye. Winona Lake, Ind.: Eisenbrauns. Pp. 65-83.

Schneider, Thomas. 1996. Review of Hoch (1994). Orientalia 65: 174-77.

Speiser, Ephraim Avigdor. 1942. "The Shibboleth Incident (Judges 12:6)." Bulletin of the American Schools of Oriental Research 85: 10-13.

Steiner, Richard C. 2000. "Semitic Names for Utensils in the Demotic Word-List from Tebtunis." Journal of Near Eastern Studies 59: 191-94.

Stevenson, C. H. 1970. The Spanish Language Today. London: Hutchinson.

Swiggers, Pierre. 1981. "The Word Sibbolet in Jud. XII.6." Journal of Semitic Studies 26: 205-7.

von Soden, Wolfram. 1965-1981. Akkadisches Handworterbuch. 3 vols. Wiesbaden: Otto Harrassowitz.

________. 1995. Grundriss der akkadischen Grammatik. 3rd ed. Rome: Pontificium Institutum Biblicum.

Vittmann, Gunter. 1996. "Semitisches Sprachgut im Demotischen." Wiener Zeitschrift fur die Kunde des Morgenlandes 86: 435-77.

________. 1997. Review of Hoch (1994). Wiener Zeitschrift fur die Kunde des Morgenlandes 87: 277-78.

Walters, C. C. 1972 [1976]. An Elementary Coptic Grammar of the Sahidic Dialect. Oxford: B. H. Blackwell.

Ward, William A. 1996. "A New Look at Semitic Personal Names and Loanwords in Egyptian." Chronique d'Egypte 71/141: 17-47.

Woodhouse, Robert. 2002. "Hebrew Sibbolet 'Ear of Grain; (Olive) Branch' and 'Stream, Torrent, Flood': An Etymological Appraisal." Studia Etymologica Cracoviensia 7: 172-89.


COPYRIGHT 2003 American Oriental Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Woodhouse, Robert
Publication:The Journal of the American Oriental Society
Geographic Code:7EGYP
Date:Apr 1, 2003
Previous Article:The Resources of History: Tradition, Narration and Nation in South Asia.
Next Article:Embracing Inana: legitimation and mediation in the ancient Mesopotamian sacred marriage hymn Iddin-Dagan A.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters