Lecture I, 50
Lecture I, 50
Abstract and Keywords
In Lecture 49, the chapter begins the discussion of the cases by drawing attention to indeclinable nouns and defective nouns (nouns lacking one or more case-forms),and by considering the relation between case-forms and adverbs. The chapter proceeds (Lecture 50) to the question of the number of cases to be found in Indo-European and non-Indo-European languages, and to the diachronic phenomenon of syncretism (reduction over time in the number of cases) and reasons for it. Postponed to a later lecture (one never published) is the treatment of the use of the individual cases, with the exception of the vocative, which, together with the use of nominative for vocative, and some remarks on the use of the vocative particles Gk □ and Lat. ο, is the subject of the last lecture of the first series (Lecture 51).
How many cases are there? Well, usually we recognize six in Latin, five in Greek, and four in modern German. The question arises how this disagreement is to be explained, and which of the three languages has best preserved the original state of affairs. In an earlier lecture, when I was trying to give an outline history of syntactic studies, I alluded to a curious statement of Gottfried HERMANN, to the effect that, from the point of view of Latin grammar, there simply could not have been either more or fewer than six cases (I, 29 above). Now, this theory, however typical of grammatical thinking of the time, is obviously wrong. We now know that the Indo‐European parent language, from which Greek and Latin derive, definitely had more cases than Latin, at least eight in fact. Alongside the cases recognized in Latin, Indo‐European certainly had an instrumental (or sociative) and a locative.1 Indeed, when we look further afield, we find a number of languages with an even richer set of case‐endings. | Finnish has sixteen cases, and there are Caucasian languages with twenty‐four distinct case forms.2 Beside these, however, we find languages with no nominal inflection at all, such as Chinese, which in its structure is not (as was once supposed) a very archaic language but very highly developed, indeed much more modern than the most modern European languages such as English. We can say that the movement in language change is to shift gradually down from a large number of case‐forms, with a constant reduction in the number of inflectional forms.3 We can see this (p.378) also in the languages which concern us here, as e.g. modern Greek shows a drastic loss of cases, and so, too, do the Romance languages.4
We must now try to clarify by what processes this simplification occurred. In discussions of this question, there is one particularly key term: syncretism. If we show, for example, that the German dative has added to the functions of the old dative those of the ablative, locative, and instrumental, then we say it is a ‘syncretistic case’. The best treatment of this subject is DELBRÜCK's monograph (1907) entitled ‘Syncretism: A Contribution to the Study of the Cases in Germanic’.5 By ‘syncretism’ one means the merging of different cases as one. Naturally, one thinks of Gk κϵράννυμι ‘mix’ and takes the term ‘syncretistic case’ to mean a mixed case. But this etymology cannot be right, if only because a derivative of that verb should have ‐krat‐, not ‐kret‐ (cf. crater, crasis). In fact, the word συγκρητισμός, coined in late antiquity, has to do with the name of the island of Crete. The island was rich in cities (Homer calls it ἑκατόμπολις (‘with a hundred cities’) at Il. 2. 649), and these cities were constantly at loggerheads, but when danger threatened from outside, feuding cities were seen to forget their quarrels and unite. So, just as ἑλληνίζϵιν, λακωνίζϵιν meant to feel that one is a Hellene, or a Spartan, whence the abstracts ἑλληνισμός, λακωνισμός, so συγκρητίζϵιν and συγκρητισμός denoted the sense of a common identity as Cretans (cf. Plutarch, On Brotherly Love, 19 p. 490B [in vol. 6 of the Loeb edn of the Moralia]). In the sixteenth century, the term was revived for analogous phenomena to refer to the uniting of different theological and ecclesiastical movements within Protestantism against Catholicism. The word was used rather pejoratively to mean the mixing and blurring of opposites, and modern theologians use it to mean simply a mixture of religions. In the nineteenth century, this group of terms was |, extended to other fields, including medicine; POTT seems to have been the first to use it of grammatical cases—CURTIUS (1875b: 192–3) refers to POTT (1833–6: I, 22).6
As to the question how reduction of cases occurs, there are three principal factors to consider. First, a reduction reflects a sort of balancing. In Latin, for example, although we distinguish six cases, it is well known that not every stem has six distinct case‐forms in both singular and plural. This continues in part a very ancient state of affairs: for example, the fact that in the plural throughout the (p.379) nominal system dative and ablative are never distinct, both ending either in ‐īs or in ‐ibus, is an inherited feature, and we find exactly the same in Indic and Iranian.—Also very old, indeed in principle older still, is the identity in all Indo‐European languages of nom. and acc. in all neuter nouns, both sg. and pl. The reason for this, to come straight to the point, is that a noun with a neuter form was never regarded as an agent but always as affected by the action of the verb (note especially the endings of the 2nd declension); even when functioning as the grammatical subject, a neuter intrinsically denoted an object in semantic terms.7—Similarly in the dual, too, from the beginning nom. and acc. were not distinguished, and in Greek the gen. and dat. were not either.
Inherited ambiguity of this kind in certain case‐forms could lead to a weakening of the sense of opposition between the cases in question. Univocal case‐forms corresponding to polyvalent case‐forms in one of their roles received by extension the other functions of the polyvalent forms, and other univocal forms which originally signalled these other functions were lost. So, it was probably under the influence of the identity of the dat. and abl. in the plural that the Germanic dat. sg. acquired abl. meaning. The dative as heir of the abl. is seen especially clearly in Gothic, where corresponding to the Latin abl. in maior eo ‘greater than he’ (and to a genitive in Greek, of course) Wulfila uses the dat. maiza imma. The same happens in Old Irish.8 (On the abl. after a comparative, see I, 5–6 above.)
Of relevance in the present context is the fact that the roles of the Latin abl. are matched in Greek partly by the dat. and partly by the gen. The Roman grammarians already made some fine observations (I, 19–20 above) on the correspondence between the Latin abl. and the Greek dat., to which we shall return shortly (I, 305 below). As for the relation between abl. and gen., | the situation is as follows. Originally, in the singular most nouns used the same form for both gen. and abl. functions, only the o‐stems (the so‐called 2nd declension) using a separate form for each case role. This original irregularity is preserved in Sanskrit, in the oldest parts of the Avesta, and in the cuneiform inscriptions of Old Persian.9 Everywhere else, levelling occurred. In Latin, the opposition between (p.380) gen. and abl. sg. inherited in the o‐stems was extended to all other stem classes by making new abl. forms modelled on that of the o‐stems. Oscan and Umbrian went in for this, too, so the innovation is Common Italic.10 Quite independently, Young Avestan extended the abl. sg. form to all stem‐classes.—Greek went about it the opposite way. The usual pattern of most nouns in the singular was extended to all stems, in both sg. and pl., so that all inherited gen. forms acquired also abl. meaning, and the old abl. sg. form was lost, except in half‐adverbial expressions such as Delphian ϝοίκω (‘from one's household, at one's own expense’; CAUER & SCHWYZER no. 323 = Buck no. 52, C23; early 4th c. BC). Slavic and Baltic also did away with the opposition between abl. and gen., but in the o‐stems they extended the gen. function to the old abl. form,11 unlike Greek, which gave abl. meaning to the gen. form. With this genitival use of the abl. compare that of the prepositions German von and French de (both ‘from’ and ‘of’), of az in modern Persian,12 and of the ending ‐θϵν in Greek (I, 300 above).
A second cause of simplification in the case system comes more from outside, and is especially relevant to Germanic and Romance. It is well known that weakening and loss of sounds occur most frequently at the end of the word, and these phenomena lead necessarily to the falling together of originally distinct case forms, giving rise to a single form which may signal very different functions, and which may then serve as a model for other paradigms. Even ancient Greek and Latin were not immune to this sort of purely phonetic merger of case‐forms.13 The fact that e.g. the form ἡδϵι̑ς is both nom. and acc. is due in the first instance to phonetic factors: ἡδϵι̑ς as nom. arises by contraction from *ἡδέϜϵς, while ἡδϵίς as acc. strictly goes back to *ἡδένς, the ‐ϵ‐ in both cases being the same as in the dat. ἡδέσι.14 This phonetic merger of nom. and acc. pl. in nouns in ‐υς (and ‐ις) then led to the use of other nom. pl. forms in ‐ϵις also in acc. function: an acc. such as ϵὐγϵνϵι̑ς | can be explained only as an analogical (p.381) imitation.15 This contributed to the fact that in the plural in modern Greek all nominatives of the old 3rd declension function also as accusatives.16—In Latin, too, the nom. pl. in ‐ēs has a different origin from that of acc. pl. ‐ēs.17
A third point takes us back to more semantic considerations. Two forms could be used originally as synonymous in terms of substance, although regarded from different points of view, and this could lead to the obsolescence and eventual loss of one of the forms. Let us take an example. One very striking and curious agreement between Greek and Germanic is that both came to lose the form used for marking instrumental, and to use instead the case‐form serving also for locative. It seems that the explanation lies in the semantic proximity between the two case‐functions. The instrumental is used when the noun is regarded as a tool or a companion, the locative, when it is a place or a container. Now, in many utterances, it is not really clear whether something is really to be regarded as locative or instrumental. In Latin curru uehi (‘to ride in a chariot’), for example, it is clear that Latin‐speakers thought of the car as instrumental, but in German we would say rather im Wagen fahren. Both utterances convey the same substance, but the car is regarded as a tool in the Latin, but as a place in the German (and English). In keeping with this is the fact that, in word formation, derivatives denoting tools and places sometimes use the same suffixes. Note Greek words in ‐τρα, for example: ἀκέστρα means needle, the tool for patching, but μάκτρα ‘kneading‐trough’ and ϕαρέτρα ‘quiver’ denote also the places where dough is kneaded (μάσσω ‘I knead’, root μακ‐) and arrows carried (root ϕϵρ‐).18 The local meaning is yet more dominant in cases such as ὀρχήστρα ‘dancing‐floor’ and παλαίστρα ‘wrestling‐school’. So, too, with the suffix ‐τ ρον, e.g. ϕέρτρον ‘bier’ denotes that by means of which, but also that on which, a corpse is carried. This semantic relation between instrumental and locative could easily lead to the widespread generalization of one of the two forms, and the gradual obsolescence of the other.—Then the locative and dative also fell together, in circumstances which we will not discuss here.19 And since in the Italic branch of Indo‐European (p.382) the same forms served for abl. and instr., the Latin abl. as marker of means and tools came to stand as the counterpart of the Greek dative. |
This reduction in the number of case‐forms occurred in most of the Indo‐European languages, and one should add that it occurred in parallel with an increase in the use of prepositional expressions, which helped to make the distinctions for which case‐forms alone did not suffice. In languages whose history we can follow over long periods, such as Latin, a portion of these two parallel and mutually conditioned developments happens before our eyes. In a number of languages, in the end, all the old case‐forms were lost, unless they served to distinguish singular and plural. This is so in the Romance languages and in modern Persian,20 and English and Bulgarian are not far off that point.21
Striking by contrast is the conservatism of certain other languages. Even today, Lithuanian and the Slavic languages (except Bulgarian) retain all the inherited cases in both singular and plural, except that they have given up the distinction between genitive and ablative (I, 303 above).22 Armenian has retained even this distinction, and instead sacrificed a separate vocative form.23 On this point Eduard SCHWYZER (1917/20b: 166) has recently made the very acute observation that the languages just mentioned were in contact with non‐Indo‐European languages with rich case systems, Armenian with Caucasian, Baltic and Slavic with Finnish, and this contact favoured their retention of their old case systems. Indeed, Ossetic, an Iranian language, under Caucasian influence even increased the inherited number of cases to ten.24 Attention was drawn earlier to this sort of influence by MEILLET (1920a); JACOBSOHN (1924), however, rejects this sort of explanation.
(p.383) In this course of lectures, I do not have time and space to go into the details of case usage or into the question of the basic and central meaning of each individual case. We shall return later to a number of these points in connection with sentence structure and case syntax.25 For the moment, let us discuss just one grammatical case in a little more detail, one which we shall not be able to consider in later sections, and which anyway occupies a rather special position.
(1) W. is right to say‘at least eight’: a ninth case, serving to indicate a goal, variously termed‘directive’,‘terminative’,‘allative’, was proposed for Indo‐European as early as 1928; for bibliography on this, see Szemerényi (1996: 159 n. 2). For up‐to‐date overviews of the case system of Indo‐European, see Sihler (1995: 246–56), and Szemerényi (1996: 157–61), with rich bibliography.
(2) For Finnish, Sulkala& Karjalainen (1992: 208) speak of‘about 16 case forms’, although they here list only 15, the first four of which‘perform clearly defined grammatical functions’: nom., acc., gen., partitive, essive, translative, inessive, elative, illative, adessive, abl., allative, abessive, instructive and comitative—there is also the prolative (and the lative, which is largely obsolete); of the first 15, Abondolo (1998a: 157) recognizes as true cases only the first 12, on the grounds that the‘adverbial’ cases (abessive, instructive, and comitative) are‘of restricted distribution and stylistically charged’. As to the Caucasus, in Udi, for example, an endangered language of the Lezgian subgroup of Northeast Caucasian, there are four grammatical cases—absolutive, ergative, gen., dat.—plus some 21 local cases (see Harris 2002: 23; 2003: 177–81); cf. Haspelmath (1993:§7.1.2) on the formation of the 18 (4 grammatical+ 14 local) cases in Lezgian.
(3) Even in ancient Chinese, however, there is no sign of earlier case‐inflection (Peyraube 2004: 995–6). On the history of the‘morphological’ classification of languages in the 19th century, see Morpurgo Davies (1998: esp. 71–5, 212–16), with notes and further references.
(4) In fact, the reduction of the system in Greek (where nom., acc., and gen. survive in the modern language) is less drastic than in Romance (where, except in Romanian, systematic case‐marking survives only in the pronoun). On the loss of the dative in Greek, see Krebs (1887–90), Humbert (1930), Browning (1983: 36–8), and Horrocks (1997a: 49, and see Index, s.v.‘dative’). On the progressive elimination of the Latin cases in Romance, see the important articles by Coleman (1976) and Plank (1979).
(5) Surprisingly little has appeared since W. For a brief review, see Meiser (1992: 188), and for general treatments of the subject, note, apart from Meiser, Colpe (1975), and again Coleman (1976) and Plank (1979).
(6) Although he mentions neither Curtius nor Pott, Meiser (1992: 202 n. 2) infers from the absence of the term syncretism from earlier key works by Grimm, Schleicher, and Delbrück that it was not really current in linguistics before the 1890s.
(7) This view (which goes back to Bopp) has been developed (notably by Uhlenbeck and Martinet) into the hypothesis that certain features of the case‐marking of IE preserve traces of an earlier,‘ergative’ stage of the language, that is, one in which a noun in agent function, typically as subject of a transitive verb, is distinguished morphologically (in the‘ergative’ case, early IE*‐s) from a noun in patient function, either as the object of a transitive verb or as the subject of an intransitive verb (in the‘nominative’ case, early IE*‐m). Neuters, never achieving agent status, appear always in the‘nominative’. For a good recent, sceptical review of the whole theory, see Villar (1984: esp. 167–70, 188–90) with references to earlier literature.
(9) So, in the o‐stems we find gen. sg. Skt‐asya : Avestan‐ahiiā : Old Persian‐ahyā vs abl. sg. Skt‐ād: Avestan‐āt̰: Old Persian‐ā t. In all other declensions, gen. and abl. sg. are identical, as in IE, except that (as W. notes below) in Young Avestan new abl. sg. forms in‐t̰ appear, contrasting with the gen. sg., in imitation of the state of affairs in the o‐stems. See, on Sanskrit, Coulson (2006: Appendix 2); on Avestan, Hoffmann (1989), Hoffmann& Forssman (2004:§§84–6); on Old Persian, Kent (1950: 58–60) and Schmitt (2004).
(10) So, we find Old Latin abl. sg. in‐ād,‐ēd,‐īd,‐ūd (in e.g. sententiad, leged, died, couentionid, magistratud) in analogical imitation of inherited‐ōd; word‐final‐d was lost in Latin towards the end of the 3rd c. BC. Oscan attests‐ād and‐īd (in e.g. toutad‘people’, slaagid‘boundary’); word‐final‐d was lost early in Umbrian.
(11) In fact, the o‐stem gen. sg. ending of Old Church Slavonic‐a and Lithuanian‐o point not to*‐ōd but to*‐ād, of which the ā has yet to be explained. Note also that the ending corresponding in the Baltic language Old Prussian is‐as<*‐oso, a variant of the IE gen. sg. ending*‐osi̯o. See Arumaa (1985:§81); Szemerényi (1996: 183–4).
(13) Formal mergers of this kind correspond to the first of Coleman's (1976: 47)‘three degrees of syncretism’. Meiser, however, terms (1992: 188–90) this sort of phenomenon‘synemptosis’, reserving, in common with other scholars, the term‘syncretism’ for the falling together not of individual forms but of functional categories (Coleman's‘third degree’).
(14) In the acc. and dat. pl., this‐ε‐ is secondary, analogical on the nom. pl. (and gen. and ?dat. sg.): the nom. pl. originally ended in*‐eu̯‐es, the acc. and dat. pl. in*‐u‐ns and*‐u‐si. The original acc. pl. is seen in Cretanυἱύνς‘sons’ (and perh. in Homericπολυ1ς‘many’, Il. 2. 4).
(16) A few have alternative 2nd‐declension endings, nom.‐οι, acc.‐ους: e.g.μαστόροι,μαστόρους beside nom.‐acc.μάστορες, toμάστορας‘skilled workman’; see Holton et al. (1997: 49–50), and cf. Thumb (1910:§§65–6).
(17) The endings of the Latin 3rd declension represent a mixture of inherited consonant‐ and i‐stem endings. In the pl., the consonant‐stems ended nom.‐ĕs, acc.‐ēs (<*‐ens<*‐n̥s), while the i‐stems ended nom.‐ēs (<*‐ei̯‐es), acc.‐īs (<*‐i‐ns).
(19) This merger seems to be well under way, if not complete, in Mycenaean, where in the sg. dat.‐ei and loc.‐i are used interchangeably, and in the pl. the inherited loc. ending‐si already, as in the classical language, stands in both case‐roles.
(20) The standard modern descriptions of contemporary Persian (e.g. Boyle 1966:§11; Lazard 1992:§53) agree that case relations are expressed by means of particles and prepositions, but earlier accounts spoke of cases and set out paradigms (see Windfuhr 1979:§2.1.2). A number of modern Iranian dialects do seem to distinguish‘direct’ and‘oblique’ cases (see Schmitt 1989b: part 4, passim). On the notable exception of Ossetic, see below. On Romance, see n. 4, p. 378 above.
(22) Old Lithuanian (of the 16th and 17th centuries) has ten cases, the seven of IE (minus abl.)+three additional, agglutinatively formed local cases, adessive, allative, and illative, the last of which survives still in some dialects; for details see Senn (1966:§§84–5), Mathiassen (1996: 38–49). On Slavic, see Entwistle& Morison (1949),§§114–16 on Russian,§§164–5 on West Slavonic,§§198–203 on Slovene and Serbo‐Croat (South Slavonic minus Bulgarian and Macedonian, which has similarly drastically reduced the inherited case system), and, for details on individual languages, the contributors to Comrie& Corbett (1993: Index, s.v.‘case’), including Scatton (1993: 202–3, 245), who notes the survival of some cases in archaic dialects of Bulgarian in the Rhodope mountains.
(23) Classical Armenian has seven cases, the eight of IE minus the voc., although a few personal names borrowed from Greek show distinct voc. forms (Clackson 2004a: 929). The gen. and dat. (sg. and pl.) are distinguished only in the pronouns, but the abl. sg. is distinct in all declensions except the o‐stems! For details, see Godel (1975: 26–37) and Clackson (2004a: esp. 929–32) with good bibliography.
(24) Ossetic, a North‐East Iranian language spoken in the central Caucasus, is an isolated relic of the ancient Scythian‐Sarmatian languages, cut off from other Iranian languages for some 2000 years. Recent accounts reckon with nine cases in eastern Ossetic (Iron), namely nom., gen., dat., abl., allative, inessive, adessive, equative, comitative, of which the last is absent from the western dialect, Digor. See Thordarson (1989: esp. 469), with bibliography.
(25) Another forlorn hope: cf. pp. 25 n. 9 and 284 n. 2 above.