Capítulo II “Desigualdades regionales en Colombia: un análisis de los factores determinantes de
II.3. METODOLOGÍA PROPUESTA PARA EL ESTADO DE LA EFICIENCIA Y LA
In addition to the six stops discussed in the two preceding sections, Sumerian had a phonemic glottal stop (/÷/). In the second half of the third millennium, however, the glottal stop was progressively lost in more and more environments, so that by the Ur III period at the latest, it had lost its status as an independent phoneme. Since the Sumerological transliteration system is based on Sumerian in its Old Babylonian or later form, the glottal stop is generally ignored in transliterations.
It is the Sumerian script that shows most clearly that the glottal stop was at one time phon-emic in Sumerian. The value of a sound sign is, as a rule, derived from the value of that same
sign as a word sign. The sign KA, for example, is used as a sound sign for zú in the noun zú-lum ‘date’. This value as a sound sign comes from its value as a word sign for zú ‘tooth’. In this way, the values of the sound signs reflect the phonemic make-up of existing words. Now, in the Old Sumerian period, the Sumerian script generally lacked sound signs for values with an initial vowel. This suggests that Sumerian had no words with an initial vowel.
Most later sound signs for VC-values had in the Old Sumerian period a value CVC with an initial glottal stop (§2.4). The sound signs an, ab, and íb, for example, were in the Old Sume-rian period used almost exclusively at the beginning of a word as CVC-signs, expressing respectively /÷an/, /÷ab/, and /÷ib/ or /÷eb/. The Sumerian words written with these same signs must have had an initial glottal stop too: an /÷an/ ‘heaven’, ab /÷ab/ ‘window’, and íb /÷ib/
‘hip’. What is true for such VC-signs, also holds for the vowel signs. They too had originally values with an initial glottal stop. The sign A, for instance, is a word sign for a ‘water’ (actual-ly /÷aj/, see §3.8). In the Old Sumerian period, it was as a sound sign on(actual-ly used for /÷a/ and it acquired only later the additional value /a/.
Thus, we can infer an initial glottal stop in words such as an ‘heaven’ and a ‘water’. Fur-thermore, some grammatical morphemes can be shown to have had an initial glottal stop, viz.
the nominalizing suffix {÷a} (§31.2), the form {÷am} of the enclitic copula (§29.2.3), and the locative case marker {÷a} (§7.7.1).
In addition, several morphemes can be shown to have had a final glottal stop. One such morpheme, for instance, is the noun má ‘boat’. In the Old Sumerian texts from Lagash, the comitative case marker is only written da after a consonant (§7.11.1), so that the following example shows that má ‘boat’ must have a final consonant:
(3) má-da tuš-a
má =da tuš-Ø -÷a boat=COM sit -NFIN-NOM
‘who remains with the boat’ (DP 602 3:10; L; 24)
Moreover, in the Old Akkadian syllabary, the sign MÁ can be used as a CVC-sign for /ma÷/ or /ma¿/ (Hasselbach 2005: 64). These two values are obviously derived from the use of MÁ as a word sign for Sumerian má /ma÷/.
The noun gala ‘lamentation priest’ may likewise have had a final glottal stop. The vowel /e/
of the plural marker {ene} and of the ergative case marker {e} are retained after gala, whereas they contract with a preceding vowel: e.g., gala-e-ne (DP 220 4:2'; L; 24) and gala-e (Nik 1:297 1:3; L; 24). Such spellings show that gala had a final consonant. Whether this consonant was a glottal stop or rather an /h/ cannot be determined as yet.
Another candidate is the noun bala ‘turn of duty’. It has forms like bala-e (NATU 3:26 rev 29; U; 21), containing an uncontracted directive case marker {e}, and bala-a-ka (OrSP 47/49 233 4; U; 21), with an uncontracted genitive case marker {ak}. Both forms presume a final stem consonant, a glottal stop or possibly an /h/.
A last example is the verb ba ‘portion out’, which also may have had a final glottal stop.
The suffix {e} of the imperfect always contracts with a preceding vowel. The Old Sumerian verbal form e-né-ba-e (VS 14:173 1:4; L; 24) shows that the suffix {e} is retained after ba and that ba therefore must have a final consonant. This final consonant may have been a glottal stop or alternatively an /h/.
During the second half of the third millennium, the glottal stop was gradually lost. Some of the steps in this development can be documented. Before the various changes are discussed, however, one preliminary point must be stressed. During the same period Sumerian also lost the consonant /h/ (§3.4.4). There are as yet few criteria for distinguishing between a lost glottal
stop and a lost /h/. The following examples may therefore inadvertently include instances where the lost consonant was actually an /h/ instead of a glottal stop.
Already in the Old Sumerian period, the glottal stop completely assimilated to a preceding consonant. This change seems to have occurred after most if not all consonants. The spelling generally does not show this change, though, especially where word signs are used.
One example of a complete assimilation of the glottal stop to a preceding consonant comes from the introductory formula of letters:6
(4) na-e-a
(a.)na=Ø ÷a-b -÷e -e -÷a what =ABSVP-3N.OO-say:IPFV-3SG.A:IPFV-NOM
‘what he says’ (e.g., Nik 1:177 2:3; L; 24)
In the Old Akkadian period, this form is generally written na-bé-a /nabbe÷a/ (e.g., ITT 2:5758 2; L; 23), ignoring the syllable-final /b/ in the spelling. In the Ur III period, the spelling na-ab-bé-a is found (TCS 1:1 2; U; 21). The oldest spellings showing an assimilation to a preceding /b/ already date from the Old Sumerian period: h~~~~é-na-bé /h~ennabbe/ from /h~ennab÷e/ (FAOS 5/2 Luzag. 1 3:18; N; 24) and h~~~~é-na-bé-a-ka /h~ennabbe÷aka/ from /h~ennab÷e÷aka/ (Ukg. 6 4:4'; L; 24).
The oldest spellings showing an assimilation to a preceding /n/ also date from the Old Sumerian period:
(5) e-na-né-éš-a /÷ennanneš÷a/
÷i -nna -n -÷e -eš -÷a
VP-3SG.IO-3SG.A-say:PLUR-3PL-NOM
‘when they had said this to him’ (FAOS 5/2 Enšak. 1 7; N; 24)
The following are further spellings showing an assimilation of the glottal stop to a preceding /n/: in-na for in-ak (BM 18962=CT 10: pl.44 24; L; 21), bí-in-na for bí-in-ak (TCS 1:148 13-14; U; 21), dnin-líl-e-ma-an-na-áĝ for ...-ma-an-áĝ (BCT 1:63 3; D; 21), and im-mi-nu-ús for im-mi-in-ús (AUCT 2:7; D; 21).
The spellings of the nominalizing suffix {÷a} (§31.2), the enclitic copula {÷am}, (§29.2.3) and the locative case marker {÷a} (§7.7.1) provide many instances of assimilations, not only after /b/ and /n/, but also after basically all other consonants.
A variety of contractions are found where the glottal stop was between two vowels. For a number of reasons, it is as yet impossible to formulate precise rules. Firstly, the amount of data is quite small. Secondly, it is generally not known which vowels were long and which were short. Finally, the Sumerian rules of accentuation are largely unknown. The following must therefore remain a preliminary list of attested phenomena.
The sequence /a÷e/ could become /ā/ or /ē/. The verb er /÷er/ ‘go’ provides examples of the contraction of /a÷e/ to /ā/:
(6) mu ... mu-da-a-re-e-ša-a-šè /mundārešša÷še/
mu Ø -mu -n -da -÷er-eš -÷a =ak =še name VP-VENT-3SG-with-go -3PL.S/DO-NOM=GEN=TERM
‘because of ... who came with him’ (HUCA 29 p. 75:4 8-9; D; 21)
A similar form of the verb er ‘go’ is ba-a-re-éš-a-ne /bāreš÷ane/ from /ba÷ereš÷anē/ (AUCT 2:105 3; D; 21). The verb ba /ba÷/ ‘portion out’, however, shows a contraction of /a÷e/ to /ē/:
6 See §8.5 for more details on this particular verbal form.
(7) in-be6-e-éš /÷inbēš/
÷i -n -ba÷ -eš
VP-3SG.A-portion.out-3PL
‘They took this as their portions.’ (MVN 13:672 7; D; 21)
Another instance of /a÷e/ becoming /ē/ involves the modal proclitic {h~a}. Already in the Old Sumerian period, {h~a} contracts with a following /÷e/ to /h~ē/ and is then written h~~~~é (§25.4.1).
In the Old Sumerian period, a glottal stop was still retained between /a/ and /a/. The spell-ings of the enclitic copula {÷am} show this (§29.2.3). E.g.:
(8) udu gu7-a-am6
udu gu7-Ø -÷a -÷am sheep eat -NFIN-NOM-be:3N.S
‘These are consumed sheep.’ (DP 246 3:2; L; 24)
From the Old Akkadian period onwards, however, explicit spellings such as -a-am6 do not occur anymore. Instead the sign sequence A.AN, which had previously been used for writing -a-am6, has now acquired the value àm. This change in spelling is obviously due to a contrac-tion of /a÷am/ to /ām/.
There is little evidence for what happened to the glottal stop after other vowels than /a/. The forms of the enclitic copula suggest that /÷/ was lost early there. The enclitic copula of the third person singular has the form /÷am/ after the vowel /a/ but was simply /m/ after the vowels /e/, /i/, and /u/ (§29.2.3). Some forms of the noun lú */lu÷/ ‘man, person’ are also suggestive. In the Old Sumerian period, the genitive of this noun is found as lú-a (Ean. 1 rev 10:25; L; 25), representing /lu÷ak/ with retention of the glottal stop between /u/ and /a/. At the same time, the ergative is written lú (Ean. 1 9:3; L; 25), representing /lū/, the contracted form of */lu÷e/. Note that in later periods the ergative of lú can be written lú-ù (e.g., Cyl A 13:11; L; 22).
A special case is the contraction of /e÷i/ across word borders, which occurs in some proper nouns made up from finite clauses. Examples are šul-den-líl-le-su from šul den-líl-le ì-su (see
§6.8.3), mu-né-mah~~~~ from mu-né ì-mah~~~~ (see §10.5.2), and ku4-ra-né-sa6 (Amherst 119 tablet 5; L; 21) from ku4-ra-né ì-sa6 (TCTI 2:L.3965 case 3 = tablet 4; L; 21).
A contraction involving an intervocalic glottal stop is also attested involving the nominal-izing suffix {÷a}, when it occurs between a vowel and the locative case marker {÷a}. It re-mains unclear, however, whether it contracts with the preceding vowel or with the following case marker. See §27.3.3 and the end of §28.3.2 for examples and more details.
3.3. The affricates 3.3.1. The /z/
The Sumerian writing system has several signs for writing the consonant /z/ in combination with other sounds, e.g., za, zé, and zú. This consonant was pronounced as [ts], that is, as a voiceless dental or alveolar affricate. The evidence for such a pronunciation comes from loanwords and the values of certain sound signs in Sumerian and Akkadian orthography. How-ever, this evidence is complex due to sound changes in Sumerian as well as in Akkadian.
Loanwords from the Old Akkadian and earlier periods show that Sumerian /z/ was pro-nounced like Akkadian /s/. Sumerian /z/ was then borrowed into Akkadian as /s/: e.g., apsû
‘underground ocean’ (Sumerian abzu), assammu (a vessel) (Sumerian an-za-am), asû ‘phys-ician’ (Sumerian a-zu), simmanû ‘travel provisions’ (Sumerian zì-munu4), pursi33tu (a bowl)
(Sumerian bur-zi.d), and sammû ‘lyre’ (Sumerian zà-mí). In the same period, Akkadian /s/ is borrowed into Sumerian as /z/: e.g., ar-za-na (a kind of groats) (Akkadian arsānu), ba-za
‘cripple’ (Akkadian pessû), and za-ba-lum (a tree) (Akkadian supālum).
The Sumerian and Old Akkadian scribes wrote the three Old Akkadian consonants /z/, /s/, and /s/ with the sound signs for Sumerian /z/. For instance, when the Sumerian sound sign za is used in an Old Akkadian word, it can have three values, za, sà, and sa. This kind of multiva-lence is similar to what is found with the signs for Sumerian /b/, /d/, and /g/, which are used for respectively Old Akkadian /b/ and /p/, Old Akkadian /d/, /t/, and /t/, and Old Akkadian /g/, /k/, and /q/ (§3.2.2). The explanation is also similar. Sumerian lacked the Old Akkadian sounds /z/
and /s/, so that the Sumerian writing system lacked sound signs for writing those Old Akkadian sounds. This problem was solved by using the sound signs for Sumerian /z/ not only for the Old Akkadian sound /s/ (which was pronounced like Sumerian /z/) but also for the Old Akka-dian sounds /z/ and /s/ (which were phonetically similar to Old AkkaAkka-dian /s/).
Old Akkadian /z/, /s/, and /s/ reflect respectively Proto-Semitic *z, *s, and *s. As is increas-ingly acknowledged, either series, the Old Akkadian as well as the Proto-Semitic, consisted of affricates (Faber 1985; Sommerfeld 1995: 35-36; Hasselbach 2005: 137; Streck 2006). The forms and spellings of the Old Akkadian suffixes with an initial /ś/ (-śu ‘his’, -śunu ‘their’, etc.) provide crucial evidence. As we will see below (§3.4.1), Old Akkadian /ś/ was pronounced as [s]. Now, the initial /ś/ of these suffixes contracts with a preceding /z/ or /d/, /s/ or /t/, /s/ or /t/
to /ss/. Compare, e.g., the Old Akkadian form DAM-sú = aššassu ‘his wife’ of the noun aššatum ‘wife’ with the form DUMU-śu = mārūśu ‘his children’ of the noun māru ‘son, child’
(MAD 1:246 2-3 and 5-6). Furthermore, if the /ś/ of the suffix is preceded by a /d/, /t/, or /t/, the contracted form shows sometimes a spelling Vt-sV or the like, instead of the normal spelling with only /s/: e.g., à-wa-at-sú = hawāssu ‘his matter’ (FAOS 19 p. 116 Gir 3) of the noun (h)awātum ‘word, matter’. Evidence such as this shows that Old Akkadian /s/ was pronounced as [ts], /z/ as [dz], and /s/ as [ts’]. Since Sumerian /z/ was pronounced like Old Akkadian /s/, it too must have been pronounced as [ts].
At the end of the third millennium, Sumerian /z/ underwent a sound change and became voiced in certain environments, so that it came to be pronounced as [dz] instead of [ts]. This happened in the same period when the originally also voiceless Sumerian consonants /b/, /d/, and /g/ became voiced in most environments (§3.2.3). Sumerian /z/ seems to have become voiced in exactly the same environments, but we do not have sufficient data to verify that for all environments. But like /b/, /d/, and /g/, it remained voiceless in word-final position. This is shown by the late Akkadian loanword engisu ‘temple cook’ (Sumerian engiz).
After the pronunciation of Sumerian /z/ had changed from [ts] to [dz], Sumerian /z/ was not pronounced like Akkadian /s/ anymore but rather like Akkadian /z/. Later loanwords, from after the sound change, certify this: e.g., burzigallu (a large bowl) (Sumerian bur-zi-gal; cf. the early loanword pursi33tu from Sumerian bur-zi.d), guzalû ‘throne bearer’ (Sumerian gu-za-lá;
cf. the early loanword kussi’um ‘throne’ from Sumerian gu-za), zammukku ‘New Year’
(Sumerian zà-mu.k), and zadimmu ‘stone cutter’ (Sumerian zadim).
During the Old Babylonian period, Akkadian /s/ was de-affricated from [ts] to [s] in stages (Sommerfeld 1995: 35-36; Streck 2006). This change is shown by a change in spelling. De-affricated Akkadian /s/ is written with the signs for Sumerian /s/, while De-affricated Akkadian /s/
is, as before, written with the signs for Sumerian /z/. After the Old Babylonian period, Akka-dian /s/ had become de-affricated in all environments and from then on it is in all environments written with the signs for Sumerian /s/.
Akkadian /s/ seems to have remained an affricate (as /s/ in Hebrew did). In a Neo-Babylo-nian list with Egyptian names, Egyptian d` (which was probably an affricate) is written in the
cuneiform script with the signs for Akkadian /s/ (Bongenaar and Haring 1994). Also, the /s/ of the Babylonian name Nabū-kudurri-usur is rendered in the Old Persian script with the sign for /c/, an affricate pronounced as [tš] (Kent 1953: 193). Finally, note that /s/ has phonetic prop-erties which favour the retention of an affricated pronunciation (Steiner 1982: 84-89).
It is still unclear whether Akkadian /z/ was also de-affricated, like Akkadian /s/, or whether it remained an affricate, like Akkadian /s/. Its spelling, however, did not change and it con-tinued to be written with the signs for Sumerian /z/, just like Akkadian /s/. Yet, this does not necessarily prove a retained affricate pronunciation because the writing system lacked separate signs for writing [z] instead of [dz] anyway. The scribes did not have any choice.
In the Hellenistic period, both Akkadian /z/ and Sumerian /z/ were written with the letter ζ in the Greek script (Maul 1991: 106), while the Greek sound ζ was represented in the cunei-form script with the signs for Akkadian /z/ (Röllig 1960: 382). Berossos (3rd century BCE), however, provides us with the unique spelling Ξισουθρος for the Sumerian name zi-u4-sù-řá.
Here, the use of Greek /ks/ for Sumerian /z/ clearly points to a retention of the old Sumerian pronunciation as an affricate.
3.3.2. The /ř/
The Sumerian writing system uses the sign DU as a specialized sound sign with the values řá and ře6 for writing the consonant /ř/ with a following /a/ and /e/.7 This consonant was probably pronounced as [tsh], that is, as a voiceless aspirated dental or alveolar affricate. It gradually lost its status as an independent phoneme during the second half of the third millennium, merging with /d/ or /r/, or being reduced to zero (Jagersma 2000). Original /ř/ continued, however, to be written in certain words until the end of the cuneiform script, although not consistently. From the Ur III period onwards but perhaps even earlier, spellings with /ř/ are nothing more than conservative spellings and do not reflect any linguistic reality anymore.
As the consonant /ř/ did not exist as such anymore in the Old Babylonian period, the lexical lists generally ignore this sound and thus leave us in the dark as to which logographically written words used to contain /ř/. As a result, only the use of the sound signs řá and ře6 indi-cates the presence of /ř/. In this way some words with earlier /ř/ have been identified (Bauer 1975-76; Krecher 1993): e.g., bař4 ‘open’, bu.ř ‘tear out’, du6.ř ‘hill’, du7.ř ‘be perfect’, enku.ř ‘inspector of fisheries’, gu4.ř ‘bull’, keše2.ř ‘bind’, ki-en-ge.ř ‘Sumer’, ku5.ř ‘cut’, níĝ-řá.n ‘rod (a unit of length)’, ře6 ‘bring’, sù.ř ‘be far’, sukud = /sukuř/ ‘high’, šu-ku6.ř
‘fisherman’, šuku.ř ‘prebendal land’, u4-řá-bu (a bird), udu = /uřa/ ‘sheep’, and uku2.ř ‘poor (person)’. There must have been many more that have not yet been recognized as such through a lack of evidence.
The Akkadians perceived Sumerian /ř/ as Akkadian /s/, that is, as [ts] (§3.3.1). Early loanwords prove this (Wiggermann 1992: 175): šukūsu ‘subsistence (field)’ (Sumerian šuku.ř), usābu (or perhaps usāpu) (a bird) (Sumerian u4-řá-bu), kusarikku ‘bison’ (Sumerian gu4.ř-alim), nikkassu ‘acount’ (Sumerian níĝ-ka9.ř),8 and probably lamassu ‘tutelary deity’
(Sumerian lama3).9 However, the Akkadians perceived also Sumerian /z/ as Akkadian /s/
7 The symbol /ř/ is conventional. It is simply a convenient way to distinguish /ř/ from /r/ in the transliteration system.
8 Note the locative níĝ-ka9-řá (RA 9 p. 158 obv 10; U; 21).
9 An /ř/ as the final consonant of lama3 can as yet not be proven with an explicit spelling with the sign DU, but is strongly suggested by the alternation between zero and /r/ for its final consonant. E.g.: ur-dlama3-ra-ke4, which includes lama3.r followed by the genitive case
(§3.3.1). Sumerian had therefore two different sounds, viz. /z/ and /ř/, that the Akkadians interpreted as Akkadian /s/. At the same time, the Sumerians heard Akkadian /s/ as Sumerian /z/ (§3.3.1). Since Akkadian /s/ and Sumerian /z/ had roughly the same pronunciation, viz. [ts]
(§3.3.1), Sumerian /ř/ must have been a different sound from Akkadian /s/, even though it sounded like [ts] to the Akkadian ear. This relation of Sumerian /z/ and /ř/ to Akkadian /s/ is parallel to how the Sumerian plain voiceless stops (/b/, /d/, and /g/) and voiceless aspirated stops (/p/, /t/, and /k/) relate to the Akkadian voiceless stops (§3.2.2 and §3.2.3). By analogy we can infer that Sumerian /ř/ was pronounced as [tsh] and must have been a voiceless aspir-ated dental or alveolar affricate. Thus, /ř/ was the aspiraspir-ated counterpart of /z/.
The close relationship between Sumerian /z/ and /ř/ is confirmed by the reflex of /ř/ in Emesal Sumerian (§1.2.3). Consider the word udu ‘ram, sheep’, which had an earlier form /uřa/, containing the phoneme /ř/.10 Although Emesal normally has a /d/ for standard Sumerian /d/, udu /uřa/ has e-zé as its Emesal cognate. A few other Emesal words show a similar corres-pondence of Emesal /z/ with standard Sumerian /d/, but in those cases we lack the evidence to prove that the /d/ comes from an earlier /ř/: e.g., du10.g ‘sweet’ (Emesal zé-eb), dugud ‘heavy’
(Emesal zé-bé.d), and so on (Keetman 2007: 42).
Through a number of sound changes, /ř/ disappeared as an independent phoneme from the Sumerian sound system. The first change was its complete loss in syllable-final position, a fate it shared with the aspirated stops (§3.2.2). This development can be illustrated with the verb šu bař4 = ba.ř = (later) bar = ba.r ‘release (lit. “open one’s hand”)’. As early as the Old Sumrian period, the stem of this verb is often written ba, ignoring the final consonant, e.g., šu e-ma-né-ba ‘he released them (two doves) therein to it’ (Ean. 1 obv 19:16; L; 25). Later such spellings remain frequent, e.g., šu im-ma-ba ‘he released (its donkey stallions on the mares)’
(St F 4:11; L; 22), šu ba-an-ba ‘he was released’ (e.g., Touzalin Aleppo 275 6; U; 21), šu bí-ba ‘he released this’ (NG 213 32; L; 21), and so on. Such spellings alternate with forms where the stem is written bař4 or bar. The frequency of spellings with mere ba, however, shows that word-final /ř/ had already been lost by the middle of the third millennium.
In syllable-initial position, that is, before a vowel, /ř/ was not reduced to zero, but changed to /r/ in one geographical area but to /d/ in another. Thus, what happened with /ř/ before a vowel differed according to dialect. It became /r/ in standard Neo-Sumerian and in the dialects of Lagash and Umma, but it became /d/ in the dialects of Nippur, Isin, Ur, and Garshana. This change took place after the Old Akkadian period and is first attested in the texts of Gudea.
The reflex /r/ of /ř/ is the only one found in the texts from Lagash and Umma. Thus, we find h~~~~é-bu-re-ne (St K 3':19'; L; 22) as against earlier ba-ra-bux-ře6 (Ean. 1 rev. 1:23; L; 25), šu ba-ba-ra-da (Cyl B 15:7; L; 22) as against earlier šu bař4-řá (e.g. RTC 35 4:2; L; 23), h~~~~ é-sù-re (CT 3 pl.35-39 BM 21335 104; L; 21) as against archaic h~~~~é-sù-ře6 (e.g., MVN 6:64 obv. 2;
The reflex /r/ of /ř/ is the only one found in the texts from Lagash and Umma. Thus, we find h~~~~é-bu-re-ne (St K 3':19'; L; 22) as against earlier ba-ra-bux-ře6 (Ean. 1 rev. 1:23; L; 25), šu ba-ba-ra-da (Cyl B 15:7; L; 22) as against earlier šu bař4-řá (e.g. RTC 35 4:2; L; 23), h~~~~ é-sù-re (CT 3 pl.35-39 BM 21335 104; L; 21) as against archaic h~~~~é-sù-ře6 (e.g., MVN 6:64 obv. 2;