Some remarks on multiple-term case labels: the Hittite dative-locative

Some remarks on multiple-term case labels: the Hittite dative-locative Dita frantíková Charles University, Prague ABSTRACT This article discusses the labeling of case as aterminological problem, specifically that of the Hittite dative-locative. During the history of Hittite studies, linguists have proposed several options for dealing with the locative. The position of the dative, an important core case, has never been doubted. In the singular, the ending -i marks both indirect objects and nominal phrases answering the question where; there is also asmall number of paradigmatic forms referred to as “endingless locatives.” In the plural, the ending is always -aš, which also serves as the genitive. The author outlines the history of case description in general and in Hittite in particular, provides abasis for the choice of case labels, weighs the advantages and disadvantages of labeling the ‘dative-locative’ simply ‘dative’, and discusses arguments for separating the locative, with special attention given to the issue of the “endingless locative.” KEYWORDS Hittite, dative-locative, nominal morphology, case labelling, endingless locative INTRODUCTION Adescription of agrammatical system, although conducted by alinguist primarily for other linguists, eventually serves multiple purposes. As such, it cannot avoid the dualism of clear-cut headings versus detailed classification. Depending on the author of the description, the entire language system may be understood as an individual instantiation of the general notion of language, or taken on its own as aunique entity without much need to correspond to what the reader or learner might have known prior to trying to understand the grammar of the given language. In nominal case description, Western-trained linguists usually restrict themselves to using case names known from the grammar of the classical languages, even when describing languages outside the Indo-European family. Even when other possibilities are available, they are most often ignored. Yet even when one applies well-known case labels, it is still necessary to further specify each of them for the language being described. If the meaning were embedded in the term itself, no detailed definitions would be necessary. As amatter of fact, however, even alabel like “dative” or “allative” can encompass quite different functions depending on the specific language. In this article, Idiscuss the Hittite nominal case which ends in -iin the singular, referred to as the dative-locative in current scholarly literature. The choice of label provides only ageneral guide to the use of the case forms, and the question whether OPEN ACCESS 22 CHATREŠŠAR 2/2020 or not one should postulate an independent locative has long been treated as aterminological issue that does not affect our understanding of Hittite nominal morphology. In browsing through adescriptive grammar of Hittite, specifically the (currently most comprehensive) Grammar of the Hittite Language (Hoffner— Melchert 2008: 74, 257), one reads that the dative and locative have merged in Hittite. These authors oppose the claim of Starke (1977: 63) that it is possible to differentiate the two cases based on their syntactic behaviour. What facts or circumstances have brought about this disagreement, and how these Hittite cases are best to be understood and described, are the topic of this article. HISTORY OF CASE DESCRIPTION The earliest descriptions of nominal case do not come from Greek or Latin grammarians, but from Pānini’s Aṣṭādhyāyī, by far the earliest descriptive grammar of any Indo-European language. However, as this masterpiece of the ancient Indian grammatical tradition only became known in Europe in the 19th century, Greek and Latin grammar have never lost their status as the “gold standard” of linguistics in the West. Since the 13th century (and even earlier; see Hjemslev 1935: 11), there has been an interest in the search for ageneralized definition of case, as opposed to listing usages under individual labels such as the ablative of separation, ablative of source, ablative of origin, ablative of comparison, etc. The texts of the Byzantine grammarian Maximus Planudes and of the scholastic grammarians of his time aimed to arrive at general characteristics of nominal case systems, such as expressing origin (Latin genitive and ablative, as presented by Simon the Dane) as opposed to expressing destination (dative and accusative). Planudes, arepresentative of the localist theory (Robins 1993: 226), sought to distinguish the independent (nominative) and dependent (accusative, dative, genitive) cases and identify more specific characteristics of each group (Blake 2004: 35). These efforts, however, were largely forgotten and only discovered later by Western linguists. In the meantime, treatments of case were confined to describing individual cases as covering anumber of meanings. The early 20th century brought agrowing interest in generalizations, and the so-called Gesamtbedeutung found its way into the work of structuralist linguists, most famously Hjelmslev and Jakobson. Hjemslev claimed that acase covers “asingle abstract notion from which one can deduce specific uses” and is meaningful purely within the oppositions of acase system (Hjelmslev 1935: 85). Along similar lines, Jakobson distinguished between the invariant meaning of acase and its extension, meaning syntactically or lexically conditioned variants. Jakobsonʾs approach has its disadvantages, since it is still not possible to describe acase solely based on knowledge of its Gesamtbedeutung, i.e. it is still necessary to list its different functions (Blake 2004: 40). However, the importance of these generalizations was that they enabled linguists for the first time to form (and compare and contrast) sets of cases based on their features. To conclude, the recurrent part of the definition of nominal case is that case is defined as an (inflectional) type of relation of anoun to its head (Kittilä 2011: 5; see further Haspelmath 2009: 506). OPEN ACCESS DITA fRANTíkoVá 23 Traditionally, cases are divided into two groups, grammatical (accusative, dative, genitive) and spatial (locative, ablative, allative, etc.); the instrumental is also sometimes included in the latter despite being atypically non-spatial case, together with comitative, abessive, adessive and other less frequently found labels. The nominative is either listed under the first group (e.g. Haspelmath 2009: 512) or viewed as standing outside this division (e.g. Anderson 2006: 95). The vocative is usually treated separately due to its pragmatic conditioning. For other designations of the division (grammatical vs. semantic, etc.), see also Haspelmath (2009: 508). APPROPRIATE CHOICE OF CASE LABELS When describing the case system of alanguage, one typically aligns paradigms based on their function (e.g., to be able to state that “the accusative is used to express the direct object”; Blake 2004: 20). The specific expression (or form) is thus related to its particular function(s). When case syncretism is encountered in alanguage, one may either recognize two homophonous case forms or stipulate that acertain case covers the meanings expressed by more than one case elsewhere. An example is the syncretism of nominative and accusative found in Hittite nouns of neuter gender, where we choose to label the specific expression (token) ‘nom./acc.’, while at the same time acknowledging the syntactic role of the token as either nominative or accusative. Acontrasting example is the genitive and dative/locative plural, where we find the ending -aš in instances defined syntactically as either the genitive, dative or locative function of acertain noun.1 Here we do not say the token is ‘gen./dat./loc.’ case, but rather specify the case based on our understanding of the text, and not only on the ending itself. As observed by Haspelmath (2009: 510), the expression “XY case” is meaningless unless it is specified which language one has in mind. The label itself is less important than the definition. The cases could just as well be simply numbered (or, referred to by the shape of their exponents), so that aparticular language would use cases 1, 2, 3, 4, for example. If the function of each case is defined, there is no need for more elaborate labeling. However, keeping familiar labels helps with understanding the grammatical description of alanguage and comparing it to that of other languages. As such, it seems more useful to call acase “accusative” or “dative” and add adefinition of what functions are covered by the specific label. For the sake of clarity, one should not therefore simply state that “the accusative covers more/fewer functions in Hittite than in Greek,” but rather compare the Hittite accusative to the Greek accusative, both understood as distinct entities within their respective case systems. If one paradigmatic slot covers the meanings of several cases as understood from other languages, the use of multiple case labels may be beneficial (Haspelmath 2009: 511). For instance, Haspelmath notes that the dative in Turkish expresses both recipient and direction; therefore, amultiple label, dative-allative in this case, could be 1 That is, from Middle Hittite onwards, after the disappearance of the Old Hittite genitive plural ending -an (Hoffner— Melchert 2008: 73). OPEN ACCESS 24 CHATREŠŠAR 2/2020 adopted for clarity. This label, however, cannot capture the full range of meanings of the case, so alist of usages must be given anyway. Another possible treatment of case polysemy is to split acase (that is, acertain form of aparadigm) into several differently labeled entities. This option is not generally taken; to cite one example, “Miller (2001: 157–158) uses several different names for the -m case of Jamul Tiipay (“instrumental, comitative, allative”), and glosses the case differently depending on the English equivalent” (Haspelmath (2009: 6)). Because the main motivation for this step is that the case in question translates to multiple cases in another language which is taken as the basis for its description, it may be more useful to simply choose just one of the labels and again complement that with amore elaborate definition. QUANTIFYING ALANGUAGE FROM SCRATCH Based on crosslinguistic surveys of inflected languages, and leaving aside languages which mark relations by bound pronouns or word order, the following case hierarchy has been proposed (Blake 2004: 169). Note that the vocative is not considered acase for the purposes of the hierarchy. nom acc/erg gen dat loc abl/inst other This hierarchy presents auniversal prediction that if alanguage has one of the cases, it must also have the cases (or at least one of the cases, if there is achoice) listed to its left. E.g., if alanguage has the ablative or instrumental, it will also have the locative. Conversely, it is not possible for alanguage to have an instrumental without also having aseparate locative. Starting from the left side of the hierarchy, we find atwo-way case system of rectus (nominative) vs. oblique (acover term including accusative/ergative and all other usages) in many Indo-European languages, e.g. in Iranian (Bielmeier 2006). This two-case system became widespread already in the early Middle Iranian era, particularly in Middle West Iranian, and developed further in New Iranian. The nominative remains the base for the rectus and is unmarked in the singular; the accusative is mostly continued as rectus as well (although under some circumstances it shifts to oblique); the genitive-dative for the most part provides the formal base for the oblique case. The ablative and locative (and sometimes also dative, e.g. in Yaghnobi) are realized with the oblique case in combination with various postpositions, though several relics of endingless locatives survive. Athree-case system contrasting nominative, accusative, and genitive is found in several languages, most notably in Semitic languages such as Akkadian and Classical Arabic, but also in Modern Greek. Anumber of languages fall into the category of four-case systems, with nominative, accusative, genitive, and dative/oblique (see Old Irish, where the prepositional/oblique case is called dative); other Indo-European languages of this type include Ancient Greek and Modern German. OPEN ACCESS DITA fRANTíkoVá 25 In afive-way system, we find aseparate dative, leaving the fifth case for oblique/ ablative. An example of such asystem is Latin, where the ablative covers awider range of meanings than what we normally expect under the label “ablative,” meaning “direction from” (e.g. the ablative of instrument or ablative of accompaniment). Systems of six and more cases are expected to have the locative plus additional case(s), namely ablative and/or instrumental (since “[t]he only other case that can be placed on the hierarchy with any confidence is the locative”; Blake 2004: 157). To these belong the majority of Slavonic languages, Classical Armenian, and Turkic and Uralic languages. In describing acase language, it is natural to identify cases along the lines of the case hierarchy (though not necessarily intentionally), starting with the nominative and the accusative/ergative. The case of possession is referred to as the genitive. There follows the dative, denoting the target of an activity or emotion (Blake 2004: 144), which is usually unproblematic to differentiate (Dryer 1986). The word dative, Latin datīvus is atranslation of dotikē, Greek for ‘giving (case)’. In both Latin and Greek, besides giving, the label is meant to denote the non-subject element of an intransitive verb (e.g. Latin fīdere ‘trust’, Greek peíthesthai ‘obey’). In Greek, the dative also covers the functions of the locative and the instrumental, which in Latin are expressed by the ablative. As for the other cases, it is up to the individual linguist how to analyse and label the case system. One may take the formal structure as the key criterion, so that whenever adistinctive formal expression is found (case ending, adposition, orientational marker), the case is given alabel with a“popular” definition covering most of the meanings of the case. Alternatively, one may approach alanguage by examining expressions for particular semantic roles, with the result that acase can be given multiple labels, e.g. “dative-locative,” or conversely be split into two or more cases without any formal distinction, e.g. with case labels related to typically dative functions such as benefactive, destinative and affective (Haspelmath 2009: 514) or various designations of spatial cases, among which the ablative, locative, allative (destination) and perlative (path) are most frequently encountered.2 THE HITTITE SITUATION: HOW IT ALL BEGAN As has been seen, the decision to label cases lies in the hands of the author of agrammatical description. For Hittite, Bedřich Hrozný, who first identified the language as Indo-European, chose to follow the traditional declensional format in his Die Sprache der Hethiter (Hrozný 1917), categorizing nouns according to their stem form (vocalic u-, i-, a-stems; consonantal l-, r(/n-), n-, ntstems). He presented examples for primarily the nominative, the accusative, the genitive and the dative of every stem, while organizing the observed tokens into charts similar to those known from grammars of ancient Indo-European languages in the order nominative, genitive, dative, 2 E.g. for Proto-Indo-European, one reconstructs the locative, ablative, and possibly the allative, while destination was expressed by the accusative. OPEN ACCESS 26 CHATREŠŠAR 2/2020 accusative, followed by spatial cases: the locative (mistakenly given as -az in the singular, which is in fact the ablative ending) and allegedly merged ablative-instrumental (now known to be separate, ablative -az vs. instrumental -it/d). Hrozný noticed that the Hittite dative is frequently used in locative meaning and with apreposition (p.9). He described this feature as acharacteristic of the Hittite dative, while still assuming aseparate locative in -az (so also on p.9 resp.16, for the i-stem nouns). On p.26, he mentioned the genitival use of the dative (referring to p.9, where he cited the form ḫar-na-a-ú-i, also seen in the table on page 8) as an example of agenitive case originating in the ancient dative, which in his opinion has secondarily acquired genitive value. Soon after Hrozný, the description of Hittite grammar was refined (and revised) by Friedrich, Forrer, Delaporte, Barton, and other scholars.3 The first Hittite comparative grammar was published by Sturtevant in 1933, and there we witness ashift in case assignments: Sturtevant listed aseparate dative singular and merged dative-genitive plural, while giving only one spatial case, the ablative. The dative, in his opinion, corresponds to the Indo-European dative and locative and denotes place where, time when, indirect object and end of motion. According to him, datives ending in -awere specialized to express the end of motion in the earliest Hittite texts (Sturtevant 1933: 166). In the plural, he adopted the multiword label of “genitive/dative” for the common ending -aš. LATER TREATMENTS OF THE HITTITE DATIVE-LOCATIVE In 1977, Frank Starke published his research on the so-called “dimensional” cases and adverbs of Old Hittite, including the terminative, locative and the dative (1977: 46). He argued against the well-accepted notion that the dative and the locative not only share the same case ending, but also syntactically must be regarded as asingle category. In his opinion, the two are to be understood as separate categories. In situations where it is impossible to state whether we are dealing with the dative or the locative, it is the dative that covers multiple functions and can be defined only as the “leftover” of definitions of other spatial cases. He suggested that the locative should be understood as an independent syntactic category, and that only those forms whose locative function can be recognized with certainty should be understood as being in “locative case.” Because the terminative in Old Hittite was expressed by aspecific ending (-a), the case ending -i, besides reflecting dative usage, must have denoted the locative. As for the dative, he proposed the question: “does the category dative exist in Old Hittite at all?” Following others,4 he listed the respective usages: dative of indirect object; dative of goal; dativus finalis; dativus possessivus; dative of interest: dativus sympatheticus; dativus commodi and dativus incommodi; dativus judicantis; dativus ethicus; dativus auctoris. The subsequent development of opinion on the Hittite dative and locative case, augmented by the discovery of new texts and publication of new comparative re3 For alist of major works written on Hittite between 1917 and 1933, see Sturtevant (1933: 6). 4 He specifically mentioned Friedrich (1960: §§205–208) as his source. OPEN ACCESS DITA fRANTíkoVá 27 search, is summarized in the grammar of Hoffner and Melchert (2008), with abundant references to older works. The authors repeatedly stress that the dative and locative have merged in Hittite (pp. 74, 257) and oppose Starke’s claim of adifference in function: “There is no basis for the claim of Starke (1977: 63–68) of different functions according to an alleged distinction between nouns referring to persons and those referring to things…. The functional merger of the two cases is shown: (1) by use of the merged dative-locative to express ‘place from which’ with inanimate referents…; (2)use of the dative-locative to express ‘place to which’ with inanimate referents already in OH… ; (3) use of the dative-locative to express location with persons in combination with postpositions, also already in OH… ; (4) use of the dative-locative singular of the enclitic personal pronouns to refer to inanimate objects” (Hoffner— Melchert 2008: 257). They further provide adescriptive list of functions for the dative-locative: indirect object, dative of disadvantage, goal, purpose or result, location, temporal uses, units of measure and dimensions, additive-incremental. IS THERE ABETTER OPTION? AND FOR WHOM? Starke’s claim that “alle Wörter, für die oben der Lokativ belegt worden ist…bezeichnen ausschliesslich Sachen, Unbelebtes und abstrakte Begriffe” (Starke 1977: 66) can be disproved by several examples,5 such as the phrase LUGAL-iperan ‘ahead of the king’ (KBo 17.15 rev. 18) which illustrates the usage of the locative in connection with animate nouns. Is Starkeʾs insufficient data, and resulting implausible argument, abasis for dismissing the possibility of separate dative and locative cases in Hittite? Or is the decision to acknowledge one or two cases here simply atechnical issue? As stated above, the labeling of nominal cases in inflected languages is arbitrary, aresult of individuals decisions of users and authors of grammars. Ideally, however, one of the main considerations should be, and in practice is, clarity for the sake of the learner. There has never been adispute whether there exists in Hittite acase marking indirect objects, or whether it is possible to express location with asingle nominal form. The differences, from Hrozný through Sturtevant and Starke to Hoffner Melchert and others, are based on how scholars understand the notion of case labels (except possibly for Hrozný, whose decisions were naturally influenced by his preliminary grasp of Hittite grammar). The option preferred by Sturtevant for the singular is that of simplification: instead of using amultiword label, the designation “dative” was chosen. As the designation itself does not clarify the multiple usages of the case, it is followed by alisting of usages, which covers those known from other languages as locative. His inclusion of the allative ending -aunder the dative cannot be followed, as here we are dealing with adifferent ending which consistently expresses ameaning distinct from those of the dative. On the other hand, Sturtevant’s merged label “dative-genitive” for the plural does not seem to have ignited blazing enthusiasm among later specialists. One may 5 For other claims of Starke disproved by later authors, see Hoffner— Melchert (2008: 259–261). OPEN ACCESS 28 CHATREŠŠAR 2/2020 assume that the reason is purely practical: it looks neater if every case is listed the same way in the plural as in the singular. And it would be extremely impractical to call acase “genitive-dative-locative," for instance, even if this label would accurately describe those instances where insufficient context does not allow one to determine the exact case expressed by the ending -aš. Starke (1977) wished to set up aseparate locative, allative and dative for Hittite. Unfortunately, his arguments for demonstrating the independence of the locative did not hold water and were soon refuted. Nevertheless, he could have argued for aseparate locative based on terminological criteria: if it seems useful for any reason, it should be possible to add the locative to the list of Hittite cases even without identifying amorphological difference. Doing so is not usual in language descriptions, but surely not without precedent. If so, one would add “locative, ending in the singular in -ior exceptionally in zero, in the plural in -aš” to the inventory of Hittite cases. In syntactic positions where it is obvious that aform denotes location, it would suffice to state that it is alocative, not adative/locative. On the other hand, the indistinguishable instances and non-core dative usages would still remain ambiguous. THE HITTITE ENDINGLESS LOCATIVE AS ASEPARATE CATEGORY In this connection, it needs to be mentioned that there are several word forms in Hittite conventionally called endingless locatives (Hoffner— Melchert 2008: 74). These forms, listed below with references, are not necessarily “locatives” in the technical sense. The term “endingless locative” covers inherited endingless forms with both directional and locational meaning, as well as Neo-Hittite innovations. Their attestations are scarce, and the interpretation of some is ambiguous and certainly not always locational. For example, Tischler (1993: 292), although listing “der endungslose Lokativ tagan”, acknowledges in the translation both directional and locational meanings: ‘nieder, zu Boden; am Boden, zur Erde, auf der Erde.’6 In his monograph on the endingless locative, Neu (1980: 8) states that the term locative must be used with reservation, acknowledging that the endingless case forms answer both the question “where?” and “where to?” He lists seventeen entries, including the supine -wan (p.45) and the adverbs takšan ‘together, in the middle?’, karū ‘before, earlier’ and lukat ‘in the morning’ (first attested in New Hittite); kitkar ‘at the head (of)’ (with both locational and directional meaning; Neu (1980:25)); šer ‘above’, apostposition with nominal origin which continues an endingless locative (Melchert (2009: 616, 617); see examples given in Puhvel (1997: 201) for both location and goal); katter ‘unterer’, built to an adjective kattera- ‘lower, inferior; infernal’; and questionable single attestations of tapuš ‘on the side’ (p.41, KBo 13.20 7´with copy KUB 8.30 Vs. 23´) and tunakkiš (p.43, KBo 3.22 Rs. 78). Four of the alleged locatives, meḫur ‘time’, lammar ‘name’, keššar ‘hand’ and 6 The term “endingless locative” for this form is also found in more recent studies, e.g. Kloekhorst (2008: 858), Boroday— Yakubovich (2018: 3, 5); Puhvel (1997: 201) refers to it as a“suffixless locative.” OPEN ACCESS DITA fRANTíkoVá 29 ḫaddareš ‘at the (road) crossing’, according to Neu do not conform to the expected formal features of endingless locative. Of the endingless locatives, locatival meaning is secure for the forms šiwat, dagan, nepiš, É-er (*per), lamman-: — šiwat ‘on the day’, purely locational use (ši-wa-at KBo 3.22 Vs. 60, ši-wa-a-at KBo 21.49 iv 8´, ši-i-wa-at KBo 25.17 i1; Rieken 1999: 102; Neu 1980: 15; Tischler 2006: 1096); — dagan ‘to the ground; on the ground’, da-ga-a-an, also da-a-ga-an and ta-ga-an (Neu 1980: 8, with further references and examples to usages as both direction and location; for the vowel see Rieken 1999: 143, quoting Tischler 1993: 294, Melchert 1994: 30, 108, 135, Kloekhorst 2008: 858); — nepiš ‘in heaven’ is attested only once in KUB 33.111+ HT 25 8´ and is considered aNeo-Hittite innovation (Neu 1980: 40); — É-er (*per) ‘in the house’, locational use in KBo 6.4 i23 (Neu 1980: 29); — lamman ‘in/on the name’ (Hoffner— Melchert 2008: 109, n.142). Four more possible forms of endingless locatives discussed in the literature are keššar ‘in the hand’, tapuwaš ‘on the side/rib’, pippit- ‘Hab und Gut’ and ŠÀ-er (*ker) ‘in the heart’. — As for keššar, listed in Hoffner— Melchert (2008: 116, n.183), the locative function of this form is doubted by Neu (1980: 33–34). Kloekhorst (2008: 471) does not recognize an endingless form (except of course as nominative singular); the form kiššarta, elsewhere supposed to be an endingless locative plus pronoun, is said here to be an instrumental. — tapuwaš ‘on the side/rib’ (ta-a-pu-u-wa-aš in KUB 9.4 i9 in the middle of the sentence) is perhaps an endingless locative. See Hoffner— Melchert (2008: 118, n.190), Rieken (1999: 209, 236). — pippit- ‘Hab und Gut’, twice used in Arzawa letter VBoT 1 (Rieken 1999: 165, following Neu 1980: 15, 35) is understood by Rieken as possibly an endingless locative. As the preceding and following phrases contain the Sumerogram +mi (VBoT 1, 3–5), the case can only be contextually determined and as it is an isolated form, the interpretation as alocative cannot be confirmed. — As for ŠÀ-er (*ker) ‘in the heart’, Rieken (1999: 52–53) discusses the analysis of ke-er-ti-it-ta, understood by Eichner as an endingless locative and possesive pronoun -ti and by Neu (1980: 32 7.3) as a‘regular’ locative ke-er-ti-. Neu (1980: 31–33) also provides examples of the form with directional semantics. All of the lemmas occurring as ‘endingless locatives’ are also attested with the regular dative/locative ending -i. Since at least some of the latter express location, the distinction cannot be that of a“general” dative/locative ending vs. an endingless allomorph specified for locational use. Because the “endingless locative” is not the only way of expressing location, there may be other specific reasons for the choice of these OPEN ACCESS