Grammatical Change in Indo-European Languages

- AMSTERDAM tT·W~l:":RR,.c."",-.,--------------------------------tsrqponmlkjih GRAMMATICAL CHANGE IN INDO-EUROPEAN LANGUAGES STUDIES IN THE THEORY AND HISTORY OF LINGUISTIC SCIENCE General Editor E.F.K. KOERNER Zentrum fur Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, Berlin efk.koerner@rz.hu-berlin.de PAPERS PRESENTED AT THE WORKSHOP ON INDO-EUROPEAN Series IV - CURRENT ISSUES IN LINGUISTIC LINGUISTICS AT THE THEORY edcbaZYXWVUTSRQPONMLKJIHGFEDCBA XVIIITH INTERNATIONAL CONFERENCE ON HISTORICAL LINGUISTICS, MONTREAL, 2007 A d viso ry E d ito ria l B o a rd Lyle Campbell (Salt Lake City) Sheila Embleton (Toronto) Elly van Gelderen (Tempe, Ariz.) Brian D. [oseph (Columbus, Ohio) Edited by Iohn E. [oseph (Edinburgh) Manfred Krifka (Berlin) Martin Maiden (Oxford) E. Wyn Roberts (Vancouver, B.e.) VIT BUBENIK JOHNHEWSON SARAH ROSE [oseph e. Salmons (Madison, Wis.) M em o ria l U n iversity o J N ew Jo u n d la n d Volume 305 Vit Bubenik, [ohn Hewson and Sarah Rose (eds.) G ra m m a tica l C h a n g e in In d o -E u ro p ea n L a n g u a g es. JOHN BENJAMINS PUBLISHING P a p ers p resen ted a t th e w o rksh o p o n In d o -E u ro p ea n L in g u istics a t th e X V lIIth In tern a tio n a l C o n feren ce o n H isto rica l L in g u istics, M o n trea l, 2007 COMPANY • . AMSTERDAM/PHILADELPHIA The origin of the feminine gender in PIEtsrqponmlkjihgfedcbaZ An old problem in a new perspective Silvia Luraghi Università di Pavia The PIE gender system consisted of two genders, most likely animate and neuter; the earliest manifcstations of feminine gender were derivational and , which in origin derived abstract nouns. This suffix also involved the suffix *-h2edcbaZYXWVUTSRQPONMLKJIHGFEDCBA gave origin to the neuter nomìnativeaccusative plural, formerly a collective rather than a count plural. The semantic development is accompanied by morphological change: in the case of the neuter nomìnative-accusative plural, a derivational suffix became an inflectional ending, while, in the case of feminine gender, a derivational suffix became the marker of an inflectional class. The two morphological developments are different, and there is no reason for assuming that one of them implied the other. However, when discussing the semantic aspect of the change, it is generally assumed that either collective preceded feminine or the other way around. In my paper I suggest a different solution by considering that the two developments must be kept separated. Keywords: abstract, collective, feminine, gender, individuation, animacy, derivational suffix, inflectional suffix 1. Introduction In the present paper I review once again the problem nine gender, of the origìnal function its two developments, the feminine a century of the suffix Le. neuter plural ending, of research and theme vowel associated on the matter.: I wili start with some well known in current chronology of the changes in which the suffix one should give up trying to establish with more than facts, and theories about the relative *-h is involved. I will suggest that 2 a relative A survey ofthe issue can be found in I.cdo-Lcmos femi- , 2 gender. Given the limits of the paper, I cannot summarize proceed to discuss what I think is misleading l. of the Indo- European *-h and of the relation between chronology (2003). between the two 4 Silvia Luraghi The origin of the feminine gender in PIE developments, because they remained it is the originaI function separate from the very beginning, of the suffix as a marker of abstract nouns plain how it could turn into the ending of the nominative/accusative on one si de, and into an inflectional class marker and that that can exneuter plural At this stage, the inflectional ending edcbaZYXWVUTSRQPONMLKJIHGFEDCBA -a was reinterpreted as a (derivational) gen- der marker. Abstract nouns were attracted relation between Beside other problems, and gender marker on the other, into the feminine collective and abstract remains current explanations feminine: some coilective nouns happened foilowing different paths. source of the 'spread' of the feminine. gender, so the initial unexplained. imply an accidental For example, according to Tichy (1993) it ail started with the word * h 2 w id h éw eh 2 , an ancient collcctive, indicating 2. Some previous from Brugmann's ("seit langem bekannten as follows what he calls some generally accepted facts und allgemein akzeptierte a. the feminine the suffix involved was in origin a collective suffix. gender is a recent innovation; ferent. While (a) is indeed a fact, supported is not reconstructed as abstract, dency to treat abstract at describing an inflectional in some way derives of the suffix .•.- h 2 take for from the collective. were reinterpreted Thus, Tichy as feminine (i.e. before it became the inflectional ending before being reinterpreted while ending of the as the marker that connect the collective with the feminine, suffìx, or as an inflectional that Tìchy's description ending, of the development cord, in which the reconstructed fìxes is unknown elsewhere raise problems. implies demonstrative pened when the suffix was derivational, of the nominative/accusative the feminine, Litscher ending this led to the creation the -zi- declension), nouns on the model of masculine to which the word for 'woman, unattested verb as gender marker. of Konzepte als were felt as referring endings to In Section 2 we have seen various problematic co llective aspects of the aileged development > [em in in e: either one has to reconstruct agreement or to ignore that the fact that abstract nouns are associated in the Indo- European Ianguages is a consequence declension. through derivation, with feminine gender of the originaI value of the suffix * -h 2, which should not be disposed of so easìly. Besides, there is apparently other way to expIain this development that should preferably between secondarily a. but that both deveIopments also have a coilective meaning. Stages l and 2 are chronologically 2. If we give up the idea of deriving the rises, summarized Table 1. Developmcnt suffix turns into an inflectional one, preserving of Suffix '-h z suffix (non-obligatory) 2. a. neuter nouns: inflectìonal -a- suffix (nominative/accusative plural, obligatory) stems: marker of inflectional class ('theme vowel, obligatory) ii in Table l: ordered, whereas stages 2a and 2b represent meaning; b. i go back to developments: a derivational l. derivational no an expIanation of the suffix, i.e. to build derived abstract nouns, which could from the collective, a new scenario two separate coincidence, I would like to suggest that there is no relation the collective and the feminine, the originaI function feminine than through be avoided. In arder to find a new solution, as neuters (Le. the endings of nouns of the thematic Setting the issue in a new perspective of the as the marker could no Ionger be treated of new feminine 3. suf- is to reconstruct into the inflectional After these 'concepts' beings, the corresponding con- by means of derivational again because "einzelne durch Kollektiva ausgedriickte wurden" points out a major role, hap- His own solution neuter plural, but was then reinterpreted belebte Wesen aufgefasst inflectionally: either as a der- that the rise of gender "ta h ; played but concord in Indo-European. a change by which the collective suffix turned animate point, because the ten- gender. All explanations ivational work, neuter plural), while Litscher (2004) suggests that the suffìx nominative/accusative feminine the evolution that some collective nouns the suffix was still derivational became morpho- of the suffix and collective as a single value of the suffix has resulted in that the feminine (1993) assumes by an array of well known in fact, the originaI meaning collective. This is an important the fact that most attempts granted are quite dif- simply as collective, but rather, already in Brugmann's and secondarily according and as they may seem at first look, the two statements logical data, (b) hides an interpretation: (1891) hypothesis, *gW ena (*gW enh), was in origin an abstract noun from an otherwise 'give birth, which caused the suffix to be re-interpreted Fakten"): b. As uncontroversial the relatives of a dead person, and later 'widow'. In this respect, current theories do not differ much explanations Litscher (2004) summarizes origin of the to refer to feminine beings, and were the first class adjectives: marker ofinflectional class and feminine gender (obligatory) (part of) its 6 The orìgin of the feminine Silvia Luraghi b. a non-obligatory, grammatical, meaningful obligatory suffix turns into a theme vowel, i.e. a purely itern, which is also interpreted noun class (i.e. of a grammaticai and collective Indo-European my hypothesis, by showing reIate to each other, and (b) how nominal changed from a system system based on referential based on degrees (a) how ab- classification of individuation in to a gender (sex and, partly, animacy). Abstract and collective is certainIy not a new idea 2 concepts nouns are of severaI types: in particular, love in English, or they can be formed out of another the nature of PIE nouns how do we know that the semantic not the other way around? onstrate extension [rom concrete that change can proceed fice il to mention of institutions lars do. Since they cannot be internally word with a derivational was from abstract to abstract in the direction in the case of abstract nouns, (see for example Sweetser to concrete: referents languages ing through meaning, mass and collective in spite of being rnultiplex, and entity, mass and coUective nouns are less entities. For this reason, abstract = -istics (linguistica concrete referents, Recently, from noun, and the suffix has of collective nouns, without necessarily nouns that have turned pass- a new functionl into collective nouns. Below (1) P er ottenere il passaporto com pilare la relativa m odulistica. (2) M anca su questa m ateria un'adeguata 'In order to receive a passport suf- Ianguages. While such nouns in into an abstract unguistics). ones: Le. it is the suffix that has acquired not some abstract extension from nouns, similar to Eng- of the adjective can be turned to English semantic -istica. This suffix is conltalian edcbaZYXWVUTSRQPONMLKJIHGFE are some examples: (3) into nouns one must fili in the relevant handbooks.' Sconto del 20% su tutta la sanitaristica '20% discount on all sanitary forms.' m a n u a lìstica . in esposizione. ware on display' The above exarnples contain nouns formed with the suffix -isiica. They are gram- of matically as derived count nouns has thc same meaning: sìngular, but refer to a plurality, The corresponding piurai of the non- 3 structure that have no internai 2. See for example Ledo-Lernos 3· lhe structure: some im- of abstract and collective as well as mass nouns. Let us start with the latter. Mass nouns who mentions analyzed, 'On this topic there are no adequate they also indicate which concern the internai concrete meaning structure, than count pIurals and count sìngu- to coIlective, I would like to mention started to be used for the derivation change Ieading from ab- As for the relation between abstract and collective, one can observe entities, internai than count nouns, even ìf they refer to concrete corresponds formed with the suffixes -tio and nouns in the Romance in the Romance similarities, of they suffixes often come to indicate coilectives: as their narne implies, abstract lish - istic, the feminine or nobility (names actions (e.g. ingressus 'the action of entering', 'the beginning ltalian ingresso 'entrance' because but to a mass (exactly as mass nouns), As an example of a suffix that is currentIy undergoing abstract suffixes, there is plenty of from abstract semantic as is the case of Latin deverbal nouns Latin only indicate portant suf- to collective, and is well attested, as in the case of action nouns turning -tus and the corresponding an activity'), as change seem to dern- English nouns such as police, adm inistration, stract to concrete a collection again are continuous, nouns rank Iower on a scale of individuation or cIasses that also indicate their members). In generai, of individuals, to count plurals: i.e. they indicate nected with -istico, which derives relational adjectives formed with the suffix ,..- h2" But Indeed, some studies of semantic that change proceeds of 'result' mechanisms they can be non-derived, 1990). However, in the specific case of abstract/collective evidence referents 7 entities, too, have a low degree of concrcteness. change. fu. This was apparently entities, coilective do not indicate a single, well individuated However, it can be useful to discuss the issue, in order to understand Abstract uniplex discrete, Because they do not present an analyzable abstract That abstract and collective are neighboring of semantic The same is true of collectives, as opposed do not refer to a collection concrete 4- continuous." while count plural nouns are multiplex separate gender). In what folIows, I wili try to substantiate stract as the marker of a multiplex gender in PIE indicate (2003: 123-125), with some examples from Latin. a passage from Tacitus. Examples of several Indo-European (1986). P er ottenere il passaporto com pilare i relativi m oduli. (2)' M ancano su questa m ateria adeguati m anuali. (3)' Sconto de/20% su tutti i sanitari in esposizione. in the terms of TaImy (2000) they are already existed in Latin, but only al a late epoch, see Fruyt (1996), thìs change are given in Nussbaurn (l)' suffixcs thar attest to 4. Talmy (2000) indicates plexity and continuity among possible fcatures of referents. Plexìry refers to the presence of units: one unit referents are uniplex, other referents are multiplex. MultipI ex referents, by an arnorphous in their turri, can be conceived rnuss (coutinuous). as formed by separate units (discontinuous) or -r- 8 'Ihe origin of the feminine gender in PIE Silvia Luraghi -istica in the examples are only collective: in other words, Note that the nouns with edcbaZYXWVUTSRQPONMLKJIHGFEDCBA Nouns there are no abstract nouns m o d u listica , where they are syntactically m a n u a listica and sa n ita ristica . not the case that some abstract nouns have turned nected with the meaning into collective for reasons con- of the base of derivation; which has acquired a new meaning, thìnk, should also be reconstructed namely on the contrary, that of concrete as a parallel Again, it is it is the suffix collective. to the semantic This, I development of * -h 2 in its sernantic extensìon from abstract to collective. After thc extension to collective, the suffix * -h 2 carne to have two quite differ- the suffix ent functions: a. beings are frequently topics. Note however that nouns with human patients: collective: used with concrete that cannot further inanimate be analyzed: it indicates value (the same referents abstract: derives action nouns function is classificatory. the inflection in other words, human a feature of referents that could mediary, meanings, nouns, took two different paths obligatory, it was integrated which did not have a count plural, what we know from the Indo-European plural neuter. From (b) it developed languages into the marker into and became as the nominative/accusative of a noun class, as we will see in the next section. referents entities direct objects/patients, entities are acted upon by human manipuJated that the PIE gender system was at an early stage a system of nomi- nai dassification Ostrowsky based on different degrees of individuation (1985)5 I would lìke to elaborate status of abstract on this idea, focusing 011 On an individuation scale, the entities that rank higher (Le. the most individubeings, which have the following features: intentionality/rationality: human beings are thought to act voluntarily, based on rational thought: b. control: human c. discourse other entities is that they can bc extent of their participation in events, the scale of individuation. However, not all inanimate entities correspond to the above description. For natural forces can bring about events, even if they do not act volutarily/ example, rationally: they can be manipulated acted upon or used by human to have an in-between ties are the referents only to a limited extent, so they usually are not beings. Among inanimate status berween human referents that can be said beings and inanimate of abstract nouns. Consider concrete for example emotions: enti- emotions beings, not only because they cannot be manipulated forces can bel, but also because beings to act. The following sentences represent some beings about emotions: (4) H is a m b itio n (5 ) S h e is d evo u red (6 ) F a ith ca n m o ve m o u n ta in s. L ed h im to d isa ster. b y en vy. Another frequent typc of abstract nouns is represented by action nouns (which afe precisely the group of nouns typically ferninine in the Indo-European Action actions), nouns refer to events (not necessarily tences in which their effects on human languages). and often occur in sen- beings are focused, such as: beings can cause a state of affairs to come into beìng, prominence: human inanimate as beings, not only as the goal of their activity, but a typical feature of inanimate nouns rank low sentences topics. Usually, inanimate on the special referents. ated ones) are human a. was first suggestcd by by the as causees, or inter- entities often occur in transitive by agents. Given the limited typical beliefs of human 'Ihe bypothesis objects/ as shown cannot exert control and bring about states of af- and most often are not discourse they typically cause human and individuation versatile, on the other hand are much less versatile: typically, they (not even to the limited extent to which natural Noun classificatìon are also frequently beings are extremely cannot move, and consequently are very relevant for human 5, sentences, or even instruments. Inanimate often also as instruments: a type of referent: the in transitive agents, besides being often fact that they can even be caused to act, i.e. they can function inanimate Frorn (a) it became increasingly of neuter a mass, a plurality could be singular): from verbs; it indicates At this point, the suffix, with its two different of developrnent. nouns to indicate found subjects and semantically fairs. Nouns that refer to inanimate also have another b. that refer to human beings, because human beings they are frequent topics are especially of discourse, interested in as opposed to entities (and animals). (7 ) V icto ry in th e p revio u s (8 ) A g o o d w a /k w ill keep yo u h ea lth y. (9 ) In jla tio n ca u sed m a n y gam e en co u ra g ed th e h o m e tea m . to g o o u t o J b u sin ess. Abstract referents are not as versatile as human beings, but they are certainly much 5. Throughout Corbctt (1991). the paper, J use the ter ms "noun class" anù "gender" as synonyms, following more Iikely to appear in a variety of states of affairs, playing different inanimate concrete entities. 'Ihey are also much more roles, than likely to bc ch osen as 9 ~ lO The origin of the feminine gender in PIE Silvia Luraghi discourse topics. rational, Action nouns so their possible of affairs in which er cases, abstract often imply a human agency derives the corresponding nouns and ferninme, of the state changed, intentional nature whose causes are generally beings forces, as is the case ofedcbaZYXWVUTSRQPONMLKJIHGFEDCBA in jìa tio n in (9).6 For al! these reasons, some abstract rank higher for this reason, than concrete they look very much felt by human nouns while the gender of non-individuated entities remained virtually un- as shown in Table 3:7 verb can occur, as in (7) and (8). In oth- can refer to events as unclear: activity, from the agentive inanimate nouns on like natural the scale of Table 3. Late Proto-Indo-European l GENDER GENDER (HUMAN) gender system Il (ABSTRACT) masculine feminine 7. of PIE gender: III (CONCRETE GENDER INANIMATE) neuter individuation. Vie are now in a position to sketch a reconstruction tern, based on the above remarks and inanimate concrete regarding of the PIE noun class sys- the individuation of human, The pre-prehistory the two gender system abstract, referents as in Table 2: It is commonly held that the systern outlined in Table 3 originated two gender system, in which the relevant feature for assignment Table 2 Prototypical features of Proto-Indo-European gendcrs class was animacy. In this originai system, the animate must be understood as prototypical categories I HUMAN Il concrete abstract, non-manipulated concrete, manipulated even if they did not share ali prototypical +intentional - intentional - intentional +control + control moderately tuted by generi c words for animals, II1 ABSTRACT INANIMATE cm';CRETE from an older of nouns to either and the inanimate (as virtually ali genders gender in ali gen- der systems), which means that some entities could be assigned to a certain gender highly individuated - control ìndìvìduated dren, which were assigned non-individuated plural count inanimate collective tionality From individuation as noun to gender Corbett The suffix interpreted position * -h 2, in origin a derivational as the marker scale. Through to nouns to adjectives point, from the morphological of inflectional The last developrnent switched genders, which contained ents, were reintcrpreted 6. That economics connected was re- in the thematic declension, which followed the thematic the marker class started to on account of the sarne vow- declension. At this point of view, the system looked very much like the paradigms stems, -il- stems, and athematic systern, which it became of the newly created long stem vowel: consequently, el was extended system its extension, class (a theme vowel). Nouns in this inflectional look more and more similar the occurrence nouns, and/or that we can reconstruct for late PIE, with -0- declension. highly individuated as indicating to referential gender. and moderately the two referential the noun class Possibly, nouns could be assigned as well. The protoypical classes must be kept in mind, because according about by a sub-distinction nearly al! genders to varying suffix within the inanimate moderately what residue ac- of individuation, individuated was brought gender: in other words, the inaniconcrete, and morphologically least individumarked by the *-h2 . When the system changed again, and switched to gender assignment based on referential human contain in 'Iable 3, in which three noun classes degrees mate gender split into two genders or classes, inanimate atcd, and abstract, to ci- nature of such categories opacity at ali stages. The three gender system reconstructed were kept distinct, to (such as ra- gender, the ferninine female referents, gender was enlarged which previously referents), ents. The three stage development belonged together is outlined by words that had to the class ofhuman with nouns with human beings male refer- in Table 4: The first two individuated genders to procreate). chil- gender, not because they referred (1991: 13) calls 'sernantic residue." The existence of a semantic (i.e. highly individuated was a change in the feature that organized from individuation capability on formaI grounds counts for possible gender assignment of a noun class, the class of nouns which had a middle on the individuation of an inflectional suffix that formed abstract to the inanimate case is consti- beings including entities, but because they lacked many features of animacy ther gender 6. features. A well known or for young animate masculine is held to escape hurnan control is clearly shown by metaphors referand cornrnonly with [ree market, starting with Adarn Smith's 'invisible hand' (i.e. the hand olGod). 7. This system is also reconstructed S. In Corbett's signed according delinition, in Matasovié (2004). "The sernantic residue cornprises to a positive semantic criterio n" (1991: 13). nouns whose gender is not as- Il ,.......- 12 Silvia Luraghi The origin of the feminine gender in PIE Table 4. the Indo- European gender system: summary of the evolution STAGE 1 ANIMATE INANIMATE STAGE 2 HUMAN ABSTRACT STAGE 3 2. it avoids resorting to the chance factor and explaining inine gender as due to the accidental 3. NEUTER rather than advocating plains the evolution As ìs well known, a two gender system is attested in the Anatolian European, lost the feminine gender or whether inine gender was established. it split from PIE before the fem- (2004) and Luraghi (forthcoming) > co llective, my hypothesis with no need to reconstruct for the Hittite data, without hypothetical languag- further, accommodates the Hittite data more or less ad ho c, changes. Brugmann, KarL 1891. Zur Frage der Entstehung G eschichte trage) 15.523-531. Ledo-Lernos, In the above sections, I have offered a new explanation gender system. In my explanation, for the rise of the PIE three I have suggested that the suffix nally served the purpose of deriving abstract nouns, underwent es: on the one hand, it was reinterpreted grammaticalized two different chang- as a collective SUffLX,and as such it became as a case ending, the nominative/accusative neuter plural. On the other hand, it became the marker of a noun class, in a threefold based on the degree of individuation human (highly individuated), animate (least individuated). marker of the -a - declension. *-112, which origi- noun class system, of referents. Noun classes in this system were abstract (moderately individuated), and concrete in- The suffix then became a themc vowel, and as such the Latcr, the relcvant feature for assignment the three noun classes changed from individuation to referential gender, thus yield- ing the familiar three gender system, with masculine, With respect to earlier explanations, into one of in this paper has several advantages: 1. it does not raise the problem opment of the a gender marker of the chronological *-112 suffix into an inflcctional marker: relation between one does not have to posit the derivation from an inflectional could serve the purpose ending, of agreement; the devel- ending and its reinterpretation or to suppose as of the gender that a derivational G ram m atical suffix B eitrdge zur Unìversiry Presso aJ L a tin ed. by H. Rosén. lnnsbruck: der Universitat. G enus. A Study G en d er. Milnchen/Newcastle: aJ the O rig in s o ] the ln d o -E u ro p ea n LINCOM Europa. Litseher, Roland. 2004. Die Genese des dritten Genus: ein neuer Versuch. Paper read at the 12. F achtagung der lndogerrnanischen Luraghi, Silvia, forthcoming. P roceedings of the 20th A nnual Matasovic, Ranko, 2004. G ender Nussbaurn, G esellschaJt, UCLA G ram m atische Indo-E uropean in Indo-E uropean. Alan. 1986. H ead and hom Ostrowskì, Manfred. Krakau, 11-16 Oetober 2004. "In do- European norninal classification: from abstract to femìninc" 1985. Zur Entstehung K ategorien, F unktion C onference. Heidelberg: in Indo-E uropean. und G eschichte Winters. Berlin/New und Entwicklung York: 'Nalter de Gruyter. des indogcrmanischen ed. by B. Sehlerath. Neutrurns. Wiesbaden: Har- rassowitz.313-323. Sweetser, Evc. 1990. From E tym o/agy Talrny, Leonard. 2000. C ognitive to P ragm atics. Sem antics. Cambridge: Val. 1. C oncept Cambridge Structuring University Presso System s. Cambridge, Mass: MIT Presso Tichy, Eva.1993. Kollektiva, Genus femininum feminine, and neuter, the one outlined Cambridge 1996. Noms de procès en latino A spects Francisco. 2003. F em in in u m F em in in e Geschleehtes, Pauls and Wilhelm Braunes B ei- ' Cambridge: lnstitut fur Sprachen und Literaturen and condusions des grammatisehen u n d L itera tu r (Hermann der d eu tsch en Sprache Fruyt, Michelle. Summary any References Corbett, Greville. 1991. G ender. 8. the ne ed of reconstructing changes. the developmentedcbaZYXWVUTSRQPONMLKJIHGFEDCBA a b stra ct > [eminine from the es. Note that, having disconnected a b stra ct il ex- of a whole system; it also accounts further words or pronouns, with the creation on the dis- of abstract nouns across genders in the ancient Indo-European development 4. the decisive role of certain of the suffix in connection The limits of this paper do not allow me to discuss this matter in detail; sec Matasovié tribution classification branch ofIndo- and quite a lot of ink has been spilled in the attempt to establish wheth- er Anatolian the creation of the fem- of collective nouns with hu- man female referents; CONCRETE MASCULINE occurrence H istarische Sprachforschung 106.1-19. und relative Chronologic im Indogermanischen. 13