A Prosodic Analysis of Stress-retraction (Nesiga)
in Tiberian Hebrew:
Prosodic Structure and Optimality-theoretic Constraints
Proof of Concept
§Third and Fourth Ranks: (V)D4f, (V)D3f, D3
Vincent DeCaen & Elan Dresher
Linguistics Dept, University of Toronto
NOTES (August 2009)[1]
PRINCIPLES OBSERVED:
· One particular conjunctive, azla, among a suite of conjunctives, is quite exceptional. This accent just happens to belong to that special class of conjunctives which has a regular substitute, munach; this class is implicated in the failure of stress-retraction generally [Review Price (1990), Law of Conjuctives, for details and summary.]
· The short word versus long word distinction (DeCaen 2008) is relevant. The explanation is that length provides a buffer syllable against stressed final syllables of preceding accents; the lack of a buffer entails failure of stress-retraction.
· The beginning of the half-verse can be a blocking environment. [Review and compare Revell (1987) on the half-verse: theory and application.]
· Prosodic oddities that appear repeatedly throughout the exceptions lists:
· the specific word shape [CVCCV:CV:]
· a glottal root-consonant involved in medial syncopation
· the 2ms suffix -əkā.
Revell (1987)
#1.16
“The view that certain combinations of accents prevent the occurrence of nesiga is demonstrably invalid. No combination of accents inhibits the development of nesiga even to the extent that nesiga fails to occur with that combination in a majority of the possible cases” (p. 15).
examples:
azla + geresh (50%)[2]
C + C (51%)
also tipcha (57%)
§1. D4f dominated by D3f: (munach telisha)D4f (azla geresh)D3f 5/7
φ'
φ φ
ω ω ω ω
ger
D3 paz / Cmun / D4f
tel / C
azl / D3f ger / Deut 4:9 / HI$.F^MER
2Sam 13:5 / T.F^BO)
D2 rev / 1Sam 29:4 / WAY.O^)M:RW.
D0f ath / Num 21:33 / WAY."C"^)
Deut 3:1
D0 sil / Deut 4:33 / H:A$F^MA^(
Jud 12:6 / WAY.O^)M:RW.
Comments:
· All preceding accents are necessarily major, given the doubled-up phonological phrases.
· The regular behaviour here and elsewhere supports the proposal of introducing a fourth degree of accents: D4f.
· Num 21:33 = Deut 3:1 are unique both by (a) preceding accent and (b) word shape.
· However, the constant appearance of [CVCCV:CV:] in the exceptions lists cannot be a coincidence; we should assume word shape is the issue.
· [but why?!]
§2. VDf 14/29 (48%)[3]
φ
φ ω
ω ω
§2.1 short words: 0/14
D0sil / C
mun / VD4f
mun / D3
telisha / )AX:AR"^Y / Jer 29:2
D3f
ger / C
azla / VD3f
dar / D2f
tevir / LFT"^T / Ex 5:7
D2
rev / VD3f
mer / LFT"^T / 2K 8:19
2C 21:17
D0f
ath / VD3f
dar / W:)FKA^L / 2K 25:29
D0
sil / W:HFYW.^ / Num 35:29
W:)FKA^L / Jer 52:33
D2
revia / VD3f
mah / D2f
pashta / )AX:AR"^Y / 2C 25:14
D0f
ath / P.FRA^$ / Lam 1:13
H:AYO)MA^R / Is 45:9
W:(OLA^T / Ezra 3:4
D0
sil / )AX:AR"^Y / 1S 24:15
W:QFRF^) / Is 6:3
D0
sil / VD3f
mun / D2f
zarqa / W:$FKA^B / Num 5:13
Comment:
· In all cases of short words, there is no buffer syllable against the preceding stressed syllables under stress-retraction.
· N.B. azla would be converted to munach (Price 1990: 132-135) under stress-retraction in Jer 29:2, Ex 5:7, 2K 8:19, 2C 21:17, 2C 25:14, Lam 1:13, 1S 24:15.
· It is not a coincidence, then, that stress-retraction should fail in this environment.
§2.2 long words: 14/15
pazer / C
tel / VD4f
azla / D3f
geresh / HI$.FB:(FH / 1Sam 30:15[4]
D0f
ath / WAY.O)M:RW. / Est 6:13[5]
D0
sil / Jud 18:19[6]
D3
tel / C
azla / VD3f
dar / D2f
tevir / YI$.F^MA( / Jer 6:7
D1
zaq / WAY.O^)M:RW. / Jud 12:5
D0f
ath / VD3f
mer / WAY.O^)K:LW. / Jer 41:1
D0
sil / WAY."^C") / Ezek 16:14
D3
tel / VD3f
mah / D2f
pashta / WAY."C:)W.^ / Jer 39:4
D2
rev / HI$.F^B:(FH / Gen 21:23
D1
zaq / WF)E$.F^BA^( / 1K 2:8
D0
sil / C
darga / VD3f
mun / D2
revia / WAY.O^)M:RW. / Josh 2:14
Judg 15:13
2S 21:4
1K 1:2
WAT.F^BO) / Ezek 2:2
Comments:
· In all cases of long words, there is always a buffer syllable under stress-retraction. Stress-retraction should always obtain.
· The conjunctive azla does not convert to munach under stress-retraction on the longer words.
· There is no minimal pair for Jer 39:4, controlling for both accents and word shape; however, there is no obvious explanation for this exception. Nevertheless, in the exceptions lists, the frequency of candidates (a) undergoing syncopation and (b) that syncopation involving glottal consonants, suggests there is an additional phonological constraint at play here. See §3.2.
· Little telisha is post-positional; stress-retraction must therefore be marked by a second telisha on the stress syllable (cf. pashta). The mss. and eds. are somewhat problematic in this particular case.
§3. serving geresh (D3f): 6/19 (32%)[7]
§3.1 virtual disjunctives (VD3f) before azla serving geresh (D3f): 1/4
φ
ω φ
ω ω
VD3fmaqqef / C
azla / D3f
geresh / T"C"^) / Ex 22:5
HFYF^H / 1S 9:2
VD3f
tel / WAY.F^B") / Gen 27:33
K:BO):AKF^ / 1S 10:5
Comments:
· Generally, a virtual disjunctive in such a right-recursive prosodic phrasing is implicated in blocking stress-retraction.
· Here on azla, we again find the same distinction between short versus long words as already observed in §2; and again, the availability of the buffer syllable explains the distinction.
· The case with a 2ms suffix is an automatic exception, and so has no bearing whatsoever on the rule here.
§3.2 azla serving geresh (D3f): general case: 4/5
φ
ω ω
D2revia / C
azla / D3f
geresh / WAY.O^)M:RW. / Judg 16:5
D1
zaqef / W:SF^PAR / Lev 15:13
W:KF^TAB / Deut 24:1
QFR:)W.^ / Jer 36:9
D1
segholta / W:KF^TAB / Deut 24:3
Comment:
· Once again, the exception in Jer 36:9 involves (a) syncopation and (b) a glottal; see §2.2.
§3.3 azla serving geresh (D3f): special case at the beginning of the half-verse: 1/10
φ
#ω ω
D0fath / C
azla / D3f
geresh / WAY.FB"^) / Gen 4:3
)IM.FL:+F^H / Gen 19:20
W:KF^TAB / Deut 17:18
W:T"C"^) / Judg 9:20
WAY."D:(W.^ / Zech 11:11
D0
sil / WAY."C"^) / Gen 19:14
W.P:LIY+"^Y / Jer 44:28
WAT."C"^) / Ezek 19:14
WAY.I$.FBA^( / 1S 20:3
W:YFCF^) / Jer 31:39
Comments:
· Undoubtedly the lone exception in Deut 17:18 is directly influenced by stress-retraction in the same word W:KFTAB in Deut 24:1 and 24:3 (see §5.2).
· [check editions and mss Deut 17:18]
· [review in detail Revell’s approach to the beginning of the half-verse: theory and application]
§4. munach serving D3: 6/6
φ
#ω ω
§4.1 munach before great telisha (D3): 4/4[8]
D0fath / C
mun / D3
great
telisha / W:KF^$:LW. / Jer 6:21
WF)E$:.F^BA^( / Ezek 16:8
D0
sil / WAY.O^)M:RW. / Judg 14:18
WAT.:KO^WN"^N / 2S 7:24
§4.2 munach before pazer (D3): 2/2[9]
D0fath / C
mun / D3
pazer / N"^L:AKFH / Ex 5:3
D0
sil / HI$.F^MER / Deut 15:9
Comments:
· As a general rule, stress-retraction obtains before the major accents (e.g., D3), but is problematic before the minor accents (e.g., D3f).
· There is a prosodic distinction between §4.1 and §4.2: great telisha is found with a four-syllable word preceding the D3; whereas pazer is found with a three-syllable word preceding. This is the cut-off in the dechi study (explained in terms of branching secondary feet) at which a word is assigned a musical ga`ya or not; presumably this would not be a coincidence, then.
· [review Price on D3]
§5. additional verbs with clitics (length distinctions): 10/25
)IM-YIS.FT"^R / Jer 23:24
W:)IM-T."RF)E^H / Lev 13:57
K.I^Y-T"C"^) / Ex 22:5 / K.I^Y-YIM.FC"^) / Deut 22:22
LO^)-NF^P:LFH / Judg 18:1 / W:LO)-YI^YR:)W.^ / Jer 23:4
LO)-HF^YFH / 1S 21:7
but see Neh 13:26 / W:LO^)-YIH:YW.^ / Ezek 34:29
W:LO^)-NA(:A&F^H / 2C 35:18
LO^)-Y")FM"R / Is 62:4
W:LO)-Y")FM"^R / Jer 7:32
LO)-YI$.FMA^( / Is 60:18
LO^)-YIK.FR"^T / Jer 35:19
):A$ER-HF^YFH / Gen 13:3
):A$ER-Y"^C") / Num 33:54
):A$ER-YO^)KAL / 1S 14:24
):A$ER-NF^PAL / 2S 2:23
):A$ER-(F^&FH / 1K 2:5
):A$ER-Y"^$EB / 1K 7:8
):A$ER-)F^M:RFH / Ezek 26:2 / ):A$ER
HFY:TFH / 2C 1:11
K.A):A$ER
M:RIYTE^M / Num 27:14
)EL K.FL-):A$ER YFBO^W) / Ezek 47:9
8
[1] Acknowledgements: esp. Jim Price, Kirk Lowery, John Hobbins, ... et al.
[2] Revell (1987) n. 15, p. 16:
(1) His examples include a wider variety of constructions that have other prosodic factors in play, esp. clitics and the laws governing them.
(2) Missing from “those not showing nesiga” (?): Gen 4:3, 19:14, Jer 44:28 (?), Neh 13:26 (cf. 1S 21:7).
[3] The roughly 50/50 figure generalizes, and is determined by the statistics of word shape in Hebrew. Revell (1987) gives a general 51% (#1.16, p. 16).
[4] The concomitant external gemination indicates that stress-retraction does in fact take place here; however, tonic telisha is missing from BHS and JKeter (sic A?) editions.
N.B. in the L ms, a masora parva circule is placed where the tonic telisha should be. The explanation for the confusion is straightforward, then: the two similar looking diacritics were confused, with resulting haplography.
Weil/BHS correctly moves the circule to end of word, over the space between words (the whole three-word phrase is at issue in mp). However, only BHL correctly restores the tonic telisha.
[5] L has tonic telisha, but it is clearly pointing in the wrong direction (as if great telisha); sic BHS. BHL corrects (as BHS should have); cf. JKeter.
[6] Tonic telisha is present in L (sic BHL, JKeter), but missing for some reason from BHS edition.
[7] This is considerably lower than Revell’s figure of 50%, using a different list (1987: #1.16, n. 15).
[8] Revell (1987): #1.17, p. 16: telisha gedola (1)?
[9] Revell (1987): #1.17, p. 16: pazer (2).