Kenneth Shields Lancaster (Pennsylvania)
CDU 801.52: 809.198.7-22
ON THE ORIGIN OF THE HITTITE PARTICLE -z(a)
In the field of Hittite linguistic studies, recent scholarship has been especially interested in the nature and origin of the particles so frequently attested in this language. Among the most intriguing of these morphemes is -z(a), about which H.A. Hoffner (1969,1973) has made important observations. He points out that, although the use of this particle in verbal sentences has long been correctly identified as "Reflexivum [...], das [...] fiir alle Personen verwendet wird" (Kronasser 1956: 145), "little is known today of the rules governing the occurrence or non-occurrence of -za in the nominal sentence" (1969: 225). His research has led him to conclude that "the first and second person pronouns, when they constitute the subject of the nominal sentence, demand -za or its oblique enclitic pronoun stand-in. Stated differently, when the subject of the nominal sentence is either the speaker or his addressee(s) or both ('inclusive "we"'), the reflexive pronoun is required. When the subject of the nominal sentence is a person or object not involved in the discourse as either addressor or addressee, no need is felt for the reflexive" (1969: 230); instead, "the enclitic -as may appear for the 3rd person" (Hamp 1984: 58), cf. H-an-za-wa-za e-es 'be alive!' and ku-is-wa-ra-as a-si DUMU-as 'who is he, this child?' In this paper I want to attempt an explanation of this distribution of -z(a) in nominal sentences by exploring the possible Indo-European etyma of the particle.
It must be emphasized from the outset that a common form for the first and second persons is anomalous from an Indo-European point of view, since comparative data imply an original unity of the second and third persons in the proto-language. Thus, Erhart (1970: 113), after a detailed analysis of the verbal markers of person attested in the dialects, concludes: "[...] es bestand wohl damals noch kein Unterschied zwischen der 2. und der 3. Person." Schmalstieg (1980: 101), too, claims that "originally there was no distinction between the 2nd and 3rd person singular in the Indo-European verbal endings. Evidence of this is the identity of the 2nd and 3rd person singular endings in the following forms: Hittite preterits (-mi conjugation) e-es-ta 'was', e-ip-ta 'took', i-ya-at (beside the 2nd sg. i-ya-as) 'made'(-Af conjugation) a-sa-as-ta 'set', da-a-as 'took', da-(a-)is 'placed', tar-na-as 'put in', me-mi-is-ta 'said' [...]. Note also the Slavic 2nd and 3rd sg. aorist forms in -tb (e.g., nacg-tb 'he began') and -stb (e.g., bystb 'you were, he was'), and the identity of the Gk. 2nd and 3rd dual endings eston 'you two, they two are'." Toporov (1961: 68-70) and Adrados (1975: 538) reach identical conclusions about the lack of opposition between the second and third persons, leading them to reconstruct an original personal (first person): non-
221
personal (second/third person) opposition for Indo-European (cf. Shields 1989a: 8, 1989b: 76-77).
It has been proposed that -z(a) derives ultimately from a form in *ti. Rosenkranz (1978: 68) thus maintains that "Heth. -z wäre aus *t vor -z assibiliert und -i dann zu Schwa reduziert worden. Die Nebenform -(a)z repräsentiert eine weitere Stufe der Abschwächung." The appearance of the third person enclitic pronoun -as as a "subject stand-in" in nominal sentences suggests that -z(a) may also have some enclitic pronominal origin. In Shields (1987), I argue that the Hittite second person singular nominative pronoun zig derives from the traditionally reconstructed pronominal stem 'Heidi. Brugmann 1911: 383 and Shields 1986: 17-18; e.g., dat. Lat. ti-bl, Umbr. te-fe, ocs te-be, OP te-bbei), to which has been affixed the particle */. According to Schmalstieg (1973: 107-108) *tei was subject to monophthongization as *tlin Indo-European; and the Hittite reflex of this monophthongized etymon (*ti) was extended by -k under the influence of the first person pronouns uk, ammuk (Szemerenyi 1964: 160). The late extension of -k to *ti is demonstrated by the fact that Hieroglyphic Hittite and Lydian do not attest this development. Thus, it is possible that -z(a) represents a weak-stress variant (characteristic of enclisis) of the Hittite second person singular pronominal stem in zi-.
In order to explain the attested first person function of -z(a), I must cite another recent proposal of mine about Hittite pronominal morphology. In Shields (Forthcoming), I derive the Hittite first person singular nominative personal pronoun uk from a contamination of the deictic particles *u (cf., e.g., "1. ubi 'wo', 1. u-ti 'so', aw. u'ti, gr. e-üte 'gleichwie', ai. u-tä 'auch sogar' [...]" [Hirt 1927:11-12]) and *k(i) (cf., e.g., "Lat. ci-s\ Gmc. he-r, OE he, Goth, hi-mma, OHG hi-tumum [cf. Lat. ci-timus], Goth. hi-dre [cf. Lat. ci-tra]; Olr. ce-n, Corn, ke-n, Gaul, du-ci; Hitt. käs, ki-ssan, directly comparable to Lat. ci-s; Gk. *ky- in Ion. setos = Att. titos', Lith. sis\ OCS si; Armen. s- [radical of the 1st pers. demonstrative, 'this' hie, near the speaker, opposed to d- = near the person spoken to, 'that' iste, n- = near a third person, far from the speaker and person spoken to, 'that' ille]" [Markey 1980: 280-281]) >. Both of these deictics originally possessed "Ich-Deixis", with the original first-person signification of *k(i) being maintained in the demonstrative pronouns of Armenian and Hittite (cf. Markey 1980: 291 and Friedrich 1974: 135) 2. On the basis of Josephson's view (1979) of assibilation in Anatolian, I further propose in Shields (Forthcoming) that the ancient first person deictic/pronominal morpheme *k(i) is reflected in "the element -z- seen in
1	Forschheimer (1953: 7-8) emphasizes the close "connexion between person and deictic function (demonstration)".
2	The original first person value of *u is suggested by its use as a first person pronoun (cf., e.g., du. nom.
*we\ OCS ve, Lith. ve-du, Go. wi-f; pi. nom. *wei\ Skt. vay-am, Hitt. wes, Go. weis) and by its appearance in first person verbal endings like du. Skt. -va, Avest. -va, Go. -u, -via, OCS -ve, Lith. -va; pi. Hitt. -weni; sg. Hitt. -u(n), Lyd. -u(n), Luw. -w(i), Hier. Hitt. -wi, Toch. A -wa, B -wa. Of course, the etymological relationship between verbal suffixes and deictic/pronominal forms is well established (cf. Szemerenyi 1980: 302-303).
222 -
the paradigm of the Hittite first person plural pronoun (e.g., acc. anzas, gen. anzel, etc.)". Josephson argues that */k/ was subject to assibilation to z (= /ts/), although the process operated without full regularity. Thus, "Hitt. kittari shows no assibilation, Lyc. siyeni has it. Demonstrative ka- (neutr. Id) shows no assibilation. Cun. Luw. za~ may possibly be the result of assibilation of palatal k' [...]. [Luw] sappa- = Hitt. sip-piya- and zallati = Hitt. zallaz show full assibilation of k' in front of e > cC (1979: 102). In other words, the incomplete diffusion of this sound change across the lexicon led to the coexistence of phonologically innovative and phonologically conservative items, similar to those found in the Balto-Slavic languages resulting from the satem palatalization, 'e.g. OCS kloniti, sloniti, Lith. [...] kleivas, sleivas [...]' (Kortlandt 1978: 240)." The first person plural personal pronoun paradigm can thus attest the passage of *k(i) to z, while the first person singular personal pronoun paradigm need not. The Hittite particle -z(a) can likewise be derived from the assibilation of the element *ki, the vowel i being reduced to schwa under conditions of weak stress (cf. Rosenkranz 1978: 68).
Altough the Hittite particle -z(a) can be derived from the phonological merger of original forms in *ti and *ki, it remains to explain the existence of cognates of -z(a) in other Anatolian languages which appear to defy this theory based on the process of assibilation. Rosenkranz (1978: 67) cites "luw. -ti, hierogl. -ti, pal. -sU-zi, lyk. -ti" as etymological parallels of -z(a). It is interesting that these forms support the derivation of -z(a) from the assibilation of *ti, but it is not immediately clear how their existence permits the simultaneous derivation of -z(a) from *ki. However, Foley's recent analysis (1977: 90-106) of universal phonological processes involving assibilation allows for such a derivation. Foley (1977: 94) observes that the preferential order for the assibilation of consonants is Dd, /t/, /p/. In other words, if I\1 has begun to assibilate, then it is implied that /k/ has begun to assibilate also; but if /k/ has begun to assibilate, no such implication can be drawn for /t/3. He further notes that before Dd fully assibi-lates to /ts/, it passes through the stage /kts/ (1977: 92). Of course, he recognizes that "assibilation is most likely to occur before y [...], but after y it is most likely to occur before i" (1977:96). It is important to recall Josephson's conclusions (1979) about the irregularity of assibilation in the attested Anantolian languages. Thus, it could have been the case that as assibilation worked its way across the lexicons of the Anatolian languages, some of those languages would have shown no assibilation of the enclitic second person form derived from *ti and only partial assibilation of the enclitic first person form derived from *ki (i.e. */ktsi/). The phonological similarity between /ti/ and /ktsi/-a consequence of the latter's combination of heavy stop (/kt/) and heavy dental (/ts/) articulation - could have easily resulted in their identical orthographic representation. Therefore, the -ti of Luwian may, in fact, represent /ti/ (from /ti/) and /ktsi/ (from/ki/).
3 On the special case of /ty/, see Foley (1977: 94).
223
The attested first-second-third person value of -z(a) in verbal sentences is probably just the analogical extension of the particle to third person function in this syntactic environment, with nominal sentences preserving the original distribution. As Kro-nasser (1956: 145) emphasizes, attempts to relate reflexive -z(a) to "*sê (lat. së, got. si-k, aksl. se-bë 'sibi'" are "scheinbar unmöglich".
Numerous etymological mysteries remain in the analysis of the Hittite particle system. However, I believe that my proposal represents a reasonable, though somewhat unorthodox, explanation of Hoffner's description of the distributional properties of -z(a).
References
Adrados, Francisco R. 1975. Lingiiistica indoeuropea. Madrid: Gredos. Benveniste, Emile. 1962. Hittite et indo-européen. Paris: Adrien Maisonneuve. Brugmann, Karl. 1911. Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. Vol. D.2: Nomina. Strassburg: Karl J. Trübner. Erhart, Adolf. 1970. Studien zur indoeuropäischen Morphologie. Brno: Opera Univer-
sitatis Purkynianae Brunensis Facultas Philosophica. Foley, James. 1977. Foundations of Theoretical Phonology. Cambridge: Cambridge University Press.
Forschheimer, Paul. 1953. The Category of Person in Language. Berlin: Walter de Gruyter.
Friedrich, Johannes. 1974. Hethitisches Elementarbuch I. 2nd ed. Heidelberg: Carl Winter.
Hamp, Eric P. 1984. "Some Implications of Hittite -z(a)". Zeitschrift für vergleichende
Sprachforschung 97.58-59. Hirt, Herman. 1927. Indogermanische Grammatik. Vol. HI: Das Nomen. Heidelberg: Carl Winter.
Hoffner, Harry. A. 1969. "On the Use of Hittite -za in Nominal Sentences". Journal of Near Eastern Studies 28. 225-230.
--. 1973. "Studies of the Hittite Particles, F'. Journal of the American Oriental
Society 93.520-526.
Josephson, Folke. 1979. "Assibilation in Anatolian". Hethitisch und Indogermanisch ed. by Eric Neu & Wölfgang Meid, 91-103. Innsbruck: Institut für Sprachwissenschaft der Univ. Innsbruck. Kortland, Frederik. 1978. "I-E Palatovelars Before Résonants in Balto-Slavic". Recent Developments in Historical Phonology ed. by Jacek Fisiak, 237-243. The Hague: Mouton.
Kronasser, Heinz. 1956. Vergleichende Laut- und Formenlehre des Hethitischen. Heidelberg: Carl Winter.
224
Markey, Thomas L. 1980. "Deixis and Diathesis: The case of the Greek ¿-perfect".
Indogermanische Forschungen 85.279-297. Rosenkranz, Bernhard. 1978. Vergleichende Untersuchungen der anatolischen Sprachen. The Hague: Mouton. Schmalstieg, William R. 1973. "New Thoughts on Indo-European Phonology". Zeitschriftfür vergleichende Sprachforschung 87.99-157.
--. 1980. Indo-European Linguistics: A new synthesis. University Park, Pa.:
Penn State Press.
Shields, Kenneth. 1986. "Some Remarks about the Personal Pronouns of Indo-European". Zeitschrift für vergleichende Sprachforschung 99.10-22.
--.1987. "On the Origin of Hitt. 2nd Sg. Nom. zig". Hethitica 7.161-171.
--. 1989a. "The Origin of the Thematic Vowel". Indogermanische Forschungen
94.7-20.
--. 1989b. "The Origin of the Umbrian Perfect Suffix *-nky~". Historische
Sprachforschung 102.74-83.
--. Forthcoming. "Hittite Nom. Sg. ulč\ Historische Sprachforschung.
Szemerenyi, Oswald. 1964. Review of Benveniste (1962). Bulletin of the School of
Oriental and African Studies 27.158-162. --. 1980. Einführing in die vergleichende Sprachwissenschaft. 2nd ed. Darmstadt: Wissenschaftliche Buchgesellschaft. Toporov, V.N. 1961. "K voprosu ob evoljucii slavjanskogo i baltiskogo glagola." Vo-prosy slavjanskogo jazykoznanija 5.35-70.
Povzetek
IZVOR HETITSKEGA ČLENKA -za
Pisec domneva, daje hetitski členek -za (kije v imenskih stavkih izpričan kot 'dvojnik' osebnega zaimka prve in druge osebe), nastal po glasovnem sovpadu indoevropskih kazalnih in zaimenskih morfemov s končaje-ma *tei in *ki. Ta podmena temelji na novejših raziskavah indoevropskega oblikoslovja in na fonoloških univer-zalijah.
225