Luraghi Gender and Word Formation

8/10/2019 Luraghi Gender and Word Formation

1/40

Studies on the Collective and Feminine inIndo-European from a Diachronic and

Typological Perspective


2/40

Brills Studies in Indo-EuropeanLanguages & Linguistics

Series Editors

Craig Melchert

University of California at Los AngelesOlav HacksteinLudwig-Maximilians-Universitt Munich

Editorial Board

Jos-Luis Garca-Ramn,University of Cologne Andrew Garrett,University of California at Berkeley Stephanie Jamison, University of California at Los Angeles

Joshua T. Katz, Princeton University Alexander Lubotsky, Leiden University Alan J. Nussbaum, Cornell University

Georges-Jean Pinault,cole Pratique des Hautes tudes, Paris

Jeremy Rau,Harvard University Elisabeth Rieken,Philipps-Universitt MarburgStefan Schumacher, Vienna University

VOLUME

The titles published in this series are listed at brill.com/bsiel
http://www.brill.com/bsielhttp://www.brill.com/bsiel


3/40

Studies on the Collective andFeminine in Indo-European

from a Diachronic andTypological Perspective

Edited by

Sergio Neri and Roland Schuhmann

LEIDEN BOSTON


4/40


5/40

CONTENTS

Preface and Acknowledgements ................................................................ vii

Introduction ..................................................................................................... Sergio Neri and Roland Schuhmann

PIE feminine *-eh in Tocharian ................................................................ Hannes A. Fellner

Das andere Wort fr Frau im Urindogermanischen .......................... Jn Axel Hararson

The Gender of Abstract Noun Su xes in the Brittonic Languages Britta Irslinger

A Tale of Two Su xes: *-h - , *-ih -, and the Evolution of FeminineGender in Indo-European ....................................................................... Ronald I. Kim

Voraussetzungen fr ein feminines Genus und Implikationen frdas Kategoriensystem des frhindogermanischen Nomens ......... Roland Litscher

Zur Emergenz von-Motion und Kongruenz im Indogermanischen Rosemarie Lhr

Gender and Word Formation: The PIE Gender System inCross-Linguistic Perspective ................................................................... Silvia Luraghi

Nominal Agreement in PIE from the Areal and TypologicalPoint of View ...............................................................................................

Ranko Matasovi

PIE *-eh as an individualizing Su x and the Feminine Gender H. Craig Melchert


6/40

vi

Feminine, Abstract, Collective, Neuter Plural: Some Remarks oneach (Expanded Handout) ...................................................................... Alan J. Nussbaum

Zum anatolischen und indogermanischen Kollektivum .................... Norbert Oettinger

GenusForm und Funktion neu betrachtet ......................................... Matthias Passer

Zum Kontrastakzent und Wurzelablaut thematischer Kollektivades Urindogermanischen ......................................................................... Thomas Steer

Index of Forms .................................................................................................


7/40

PREFACE AND ACKNOWLEDGEMENTS

The idea for the present volume,Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective, was suggestedby some participants at the end of the 4. Jenaer Indogermanistisches Kollo-quium Kollektivum und Femininum: Flexion oder Wortbildung? Zum Anden-ken an Johannes Schmidt (2010) and was realized thanks to the support ofone of the lecturers, H. Craig Melchert, co-editor of the series BSIEL.

The book is presented here by request of the editors of the series as asort of monograph and is meant to exemplify the variety of approachesto the topic. For this purpose, the contributions were editorially uni ed,alphabetically arranged by authors name and provided with an Englishabstract. This allows readers who are interested just in single papers topurchase each one separately. For this reason, no comprehensive bibliog-raphy was put together for the whole volume. We would like to express our thanks to the many people and institu-

tions that made this project possible. First of all we are indebted to thespeakers at the above mentioned colloquium who, despite the short timespan between the invitation and the date of the conference and the lim-ited nancial support, were pleased to take part, as well as the numer-ous participants from near and far who have contributed to the successof the event. Without the active support of Rosemarie Lhr, the studentassistants, and the Friedrich Schiller University of Jena, the organizationof the conference would not have been possible: also to them goes ourdeep gratitude. We also are much indebted to the authors who did notactually participate at the colloquium itself, but were disposed to submita contribution to the volume. We would also like to thank the two peerreviewers of Brill for their helpful and detailed comments on the singlepapers of the book.

Finally, we want to express our heartfelt thanks to the editors of the series Brill Studies in Indo-European Languages & Linguistics, Olav Hackstein andH. Craig Melchert, who have gladly agreed to include this volume in theseries, as well as the always helpful and patient assistant editors of Brill

Franca de Kort, Jasmin Lange and Stephanie Paalvast. We are pleased to hand over the nal work to friends, colleagues andthe scienti c community.

Jena, December 2013Sergio Neri and Roland Schuhmann


8/40

GENDER AND WORD FORMATION: THE PIE GENDER SYSTEM INCROSS LINGUISTIC PERSPECTIVE

Silvia Luraghi

Abstract: The paper tackles the issue of gender as a non-prototypical category atthe border between derivation and in ection. It explores derivational and in ec-tional properties of gender, setting the PIE gender system in the wider framework

of nominal classi cation. It compares gender with other types of non-prototypi-cal categories, such as evaluation (non-protoypical derivation) and number (non-protoypical in ection). The discussion then turns to the rise and the developmentof gender systems taking agreement as their distinctive feature. Cross-linguisticdata o fer evidence for di ferent patterns of development, based on di ferentialmarking of core arguments, on the grammaticalization of classi ers or of deriva-tional a xes. Arguably, di ferential marking of core arguments is responsible forthe rise of agreement in the earlier PIE animacy-based two-gender system, whilegrammaticalization of a derivational a x provided the basis for the later sex-based three-gender system. These two di ferent origins mirror the double-faceted

nature of gender, which feeds syntax, in that it provides a means for referencetracking through agreement, and also serves the lexicon, as a means for creatingnew words and motivating groups of words. In the nal part of the paper, somepossible evidence is presented, which might indicate mechanisms that can leadto the rise of agreement through the spread of a derivational a x.

1. I

The category of gender has traditionally puzzled linguists and grammar-ians. Among its most striking features is its double-faceted nature, whichlocates it somewhere between syntax and the lexicon: gender is not in ec-tional in nouns, the lexical category for which it is most relevant, but itstrademark is its agreement triggering force, which variously manifestsitself on adjectives, pronouns, verbs, or other lexical categories associated with nouns. Far from being a peculiarity of the Indo-European languages,the double-faceted nature of gender is its distinctive feature in all lan-guages endowed with some sort of gender system. It is the purpose of

this paper to explore derivational and in ectional properties of genderas reconstructed in PIE in cross-linguistic perspective, in the light of datafrom the Indo-European languages and from other language families.


9/40

200

The paper is organized as follows. In sec. 2 I start by discussing proto-typical features of in ection and derivation, and explore the position ofgender and other categories of nouns in such a framework. In sec. 3 I godeeper into the relation between gender and other more or less proto-typical word formation processes. I then survey the features of di ferentsystems of nominal classi cation, still in connection with word formation.Sec. 4 contains a discussion of formal aspects involved in the derivationof nouns that denote female animate entities from nouns that denotemales. In this framework, I distinguish between gender motion and gen-der shift, that is derivation through the addition of a derivational a x

(not of a stem formation a x) versus derivation brought about by meansof in ectional class change (that is, involving di ferent stem formation).Sec. 5 calls attention to similarities and interplay between gender and twoother non-prototypical categories, diminutives (non-prototypical deriva-tion) and number (non-prototypical in ection). In sec. 6 I summarize thedevelopment of the PIE gender system, arguing that the change that ledthe su x *-h to become the feminine marker must be understood aschronologically disconnected from the change that let it to become theending of the nominative/accusative neuter plural. I show that attemptsto order the two changes chronologically have raised a number of issues,including the question why the su x appears throughout the paradigm infeminine nouns, while it is limited to the nominative/accusative endingsin the neuter. This and similar problems nd a solution if one conceivesof the two changes as independent developments. In sec. 7 I turn to thefunction of gender in syntax and discourse, and show that gender behaveslike in ectional categories in this respect. Sec. 8 is devoted to the possibleorigins of gender systems. It describes data from di ferent languages in

which genders came about through grammaticalization of nominal clas-si ers, of derivational a xes, or of di ferential case marking. Based onthe developments surveyed in sec. 8, in sec. 9 I elaborate on the rise ofagreement in PIE, distinguishing between the earlier animacy-based two-gender system, and the later sex-based three-gender system. In sec. 10 Idescribe some instances of noun modi er agreement through derivationala xes. Sec. 11 contains the conclusion.


10/40

- 201

2. I D M

2.1. De nitionsThe traditional distinction between in ectional and derivational mor-phology is based on a series of characteristic functions, prototypically ful-lled either by in ection or by word formation (see Dressler 1989, Dressler

et al. 1987). They are summarized in Table 1.

Table 1. Prototypical di ferences between in ection and derivation.

In ection Derivationno change of word class change of word classobligatory (=high productivity) non-obligatory (=limited productivity)syntactic function no syntactic function

Among categories of nouns, case is prototypically in ectional: in ectionfor case is obligatory for all members of the lexical classes that presentit as a relevant category (in the Indo-European languages, such lexicalclasses are noun, adjective, pronoun, and to varying extents determiner where available); in addition, two forms in ected in di ferent cases, asfor example Latinurbis(gen.) andurbe (abl.), belong to the same wordclass but bear di ferent information regarding their possible syntacticfunction.

As opposed to prototypical in ection, which serves syntax, derivationserves the lexicon, in the rst place by creating new words, which proto-typically belong to di ferent lexical classes: for example, modal adverbs are

typically made out of adjectives, while action nouns are mostly derivedfrom verbs. In the second place, derivation motivates the lexicon, as it cre-ates groups of words marked in the same way after some type of commonfeature: for example, English modal adverbs take the su x-ly, Germandeverbal action nouns take the su x-ung, and so on (see Dressler et al.1987: 99). In this sense, when it applies to nouns derivation is a type ofnominal classi cation.

These two functions of derivation (or word formation) are described as follows inDressler et al.: (a) lexical enrichment . . ., i.e. forming new words, thus serving the cogni-tive function of language . . . (b) morphotactic and semantic motivation of existing words,thus facilitating the communicative function of language . . . as well as storage in memory(1987: 99).


11/40

202

2.2. Non-Prototypicality of Gender

It has long been noticed that not all morphological categories display allthe prototypical features of either in ection or derivation. Kuryowicz(1964: 17) makes the following observations regarding number, which hecompares to case on the one hand and to prototypical derivation on theother:

If the two in ectional forms di fer semantically only, like Latinurbs (singu-lar):urbes(plural), the status of such a pair will be intermediate between therelation basic word:derivative and the relationurbs:urbem. [. . .]urbs:urbanus= two di ferent words;urbs:urbes= one word, with forms semantically di fer-ent and having secondary syntactical functions;urbs:urbem =one word, withforms semantically identical, syntactically di ferent (Kuryowicz 1964: 17).

Note further that, in the Indo-European languages, number also serves syn-tax, as it has the function of cross-referencing the subject and the nite verb.This is not necessarily so cross-linguistically, as we will see in sec. 5.2.

Gender has a further peculiarity with respect to number and othernominal categories: it is inherent in nouns, that is, it is speci ed in thelexicon. Nouns are assigned a speci c gender, and do not in ect in

the other genders available in a language. This distinguishes nominalgender from adjectival and pronominal gender, which is not inherent butin ectional. Indeed, gender of adjectives and pronouns (such as demon-stratives and anaphoras) has a syntactic function, since it either indicatesconstituency or serves for cross-reference though concord (see sec. 7).

This di ference is also captured through the distinction between inherent vs. contex-tual in ection, described as follows by Boij: Inherent in ection is the kind of in ectionthat is not required by the syntactic context, although it may have syntactic relevance.Examples are the category number for nouns, comparative and superlative degree of theadjective . . . Contextual in ection, on the other hand, is that kind of in ection that is dic-tated by syntax, such as person and number markers on verbs that agree with subjectsand/or objects, agreement markers for adjectives . . . (1996: 2). Following this approach,even some manifestations of case are not always prototypically in ectional, as so-calledsemantic cases have a meaning by themselves, and their occurrence is not triggered bythe context. As a consequence, they may have a function within word formation, whilestrictly syntactic cases may not. Regarding case, Boij remarks: [t]raditionally, a distinctionis made between structural case and semantic case. In Hungarian, for instance, nomina-tive and accusative are structural cases (i.e. contextual in ection), whereas there is also anumber of semantic cases, such as the inessive ( in ) that functions as inherent in ection. As predicted by our hypothesis that it is only inherent in ection that feeds word forma-tion, semantic cases do occur inside Hungarian complex words, unlike structural cases.(1996: 10).


12/40


13/40

204

from nouns, some of which, both from ancient and from modern Indo-European languages, are listed in Table 2. I have included instances ofderivation through gender shift to highlight the parallel with other typesof denominal nouns.

Table 2. Some denominal nouns in the Indo-European languages.

diminutives putra son putraka (Skr.) funis rope funiculus (Lat.)

Hund (m.) dog Hndchen (nt.) (Germ.) gatto cat gattino (It.)

donna ( f.) woman donnino (m.) (It.)augmentatives palla ( f.) ball pallone (m.) (It.)agent nouns mercatus market mercator (m., f.: mercatrix)

merchant (Lat.) Kunst art Knstler (m., f.: Knstlerin) artist(Germ.)science scientist (Engl.)

names of place granum grain granariusgranary (Lat.)acqua water acquaiosink (It.)

names of trees malum (n.) apple malus ( f.) apple tree (Lat.)

mela ( f.) apple melo (m.) apple tree (It.) banana ( f.) banana banano (m.) banana tree(It.)cerisecherry cerisier cherry tree (Fr.)

pomme apple pommierapple tree (Fr.)abstract nouns phsis nature phusik physics, science of nature

(Gr.)lingualanguage linguistica linguistics (It.)

collectives Berg mountain Gebirge mountains (Germ.) frutto (m. sg.) piece of fruit frutta ( f. sg.) fruit (It.)muro (m. sg.) wall mura ( f. pl.) city walls (It.)

The above list highlights similarities between gender shift and other typesof derivation, and shows that, following the de nition by Dressler andhis associates regarding the function of derivation given in sec. 2, bothgender and other noun formation devices shown above do not only createnew words, but also motivate the lexicon, as the a xes involved indicateclasses of nouns that denote classes of entities.

An important di ference lies in the degree of semantic motivation.Most frequently, gender shift is connected with referential gender (sex);

The femininemercatrix is only attested in Medieval Latin. See sec. 10.1 for more detailson the development of this su x after Classical Latin.


14/40

- 205

however, as well known, nouns with animate female referents are far fromaccounting for the whole feminine gender in most Indo-European lan-guages: the most frequent situation is one in which feminine nouns alsorefer to a large number of inanimate entities. This partly happens ran-domly, and sets gender apart from other types of derivational processes, which are better motivated semantically.

This is not to say that the classi catory power of gender is not furtherexploited, even in languages with sex-based gender systems. The classi -catory function of gender assignment to various types of inanimate nouns,as well as to nouns that refer to animals, has been described for example

for German (see Zubin/Kpke 1986); in Italian various sub-classes of inan-imate entities are consistently assigned either the feminine or the mascu-line (often hyponyms are assigned the same gender as the hyperonym, cf.Thornton 2003), and the same can be said for other gender systems. Asshown in Table 2, both in Italian and in Latin there is a systematic gendershift that motivates two noun classes, that is, names of fruits and names oftrees, whereby in Italian the former are feminine and the latter masculine, while in Latin the former are neuter and the latter feminine. Note thatthe non-correspondence between gender and type of entity across the twolanguages rules out possible semantic motivation for gender assignmentto either speci c type (as for example that trees are feminine because theybear fruits: it could seem to hold in Latin, but it does not in Italian).

Gender shift in Italian further keeps distinct several other pairs ofnouns, without allowing semantically motivated classes to be set up, asin foglio sheet of paper / foglia leaf; gobbo hunchback / gobba hump;tavolo table, desk /tavola kitchen table, table in a book (see Koch 2001:1166 for possible semantic connections between members of such lexical

pairs in Italian). Here semantic opacity is maximized, contrary to whatcommonly happens with other types of derivation. However, it must bepointed out that some opacity also characterizes other derivational proc-esses. The most typical in this respect is constituted by evaluative mor-phology: in particular, diminutives display similarities with gender in thisrespect. I will elaborate on the similarity between gender and diminutivesin sec. 5.1.

In Table 2, I have assumed that the direction of derivation goes from the noun thatdenotes the fruit to the noun that denotes the tree. Reasons of space do not allow me todiscuss this matter further here; note only that in the Indo-European languages in whicheither type of noun bears a speci c derivational a x, it is the name of the tree which isusually derived.


15/40


16/40

- 207

of gender systems (see sec. 7 on the consequences of these properties forour understanding of the nature of gender).

3.3. Nominal Classi cation and Derivation

A de nition of gender based on obligatoriness and agreement, as indi-cated above, also includes noun class systems such as that of the Bantuand other Niger-Congo languages. Similar to the sex-based gender systemsof the Indo-European languages, Bantu noun classes are obligatory (eachnoun must be assigned to a noun class in the singular and one in theplural), and they trigger agreement on adjectives and pronouns. Indeed,noun class is considered synonymous with gender in this sense (see Cor-bett 1991), and both are distinct from nominal classi ers systems, which,as remarked above, are not obligatory and do not trigger agreement. Notethat the interplay between number and gender is especially high in theBantu languages. For this reason, Grinevald and Seifart (2004) distinguishnoun class from gender: following a common practice in African linguis-tics, they use noun class for each group of nouns identi ed by a singlea x, regardless of number, and gender for the agreement classes involv-

ing the same lexical bases in the singular and in the plural. The numberof genders in this sense, i.e. classes of agreement that also account forpatterns depending on number, is clearly not the same as the number ofnoun classes as traditionally calculated.

Also in the case of Bantu noun classes, similarity with other word for-mation devices is easily shown, and has accordingly been noted by variousauthors (see for example Walter 1982: 2256), to such an extent that wordformation is sometimes regarded as the basic function of noun class pre-xes: [t]he basic function of noun classes beyond classi cation itself

imperfectly ensured in any caseis to form nouns. That noun classes playa role in derivation has naturally been recognized in all studies on thesubject. (Mufwene 1980).

Example (2) from Tswana illustrates the issue:

(2) a. cl. 1mo-sadi woman cl. 2ba-sadi women

cl. 7se-sadi feminine behavior cl. 9-tshadi group of women cl. 11lo-sadi group of women cl. 14bo-sadi womanhood b. cl. 3mo-retlwa tree of the species moretlwa


17/40

208

cl. 2me-retlwa trees of the species moretlwa cl. 9-thetlwa fruit of the moretlwa tree

cl. 10di-thetlwa fruits of the moretlwa tree cl. 11lo-retlwa thicket of moretlwa trees (from Grinevald/Seifart 2004)

Insofar as gender, even in a relatively small system such as the PIE one(three genders), or in more reduced systems (two genders) identi es atleast in part classes of entities, it ful ls a classifying function, similar tolarger systems of noun classes or to systems of non-obligatory classi ers.The relation between nominal classi cation and noun formation is cor-rectly summarized as follows in Kihm (2005):

Let me just indicate for the present that there is nothing surprising in thefact thatnoun formation, in a sense to be clari ed presently, and classi ca-tion should go hand in hand. Indeed, it belongs to the inherent propertiesof nouns that they denote entities that can (perhaps must) be allotted to dif-ferentclasses ofthings, by virtue of innate and culturally informed cognitiveprocesses, diversely expressed in languages.

4. G N -P D : F A

4.1. Interplay of Gender and Other Word Formation Processes

Table 2 also highlights the interplay between gender and other deriva-tional a xes. As is well known, some of these assign gender to the derivednoun: for example, diminutives are all neuter in German, regardless of thegender of the base (see further sec. 5.1). Agent nouns are by default mascu-line in the Indo-European languages, and if a feminine needs to be madeout of them, then some other derivational a xes must be employed. Thiskeeps apart gender shift, when indicated by in ectional class change, fromother word formation devices: the fact that gender is obligatory for nouns,and that each noun must necessarily belong to a gender, has the obviousconsequence that other word formation devices, i.e. derivational a xes, which are not obligatory, must necessarily enter some sort of relation withthe gender of the noun to which they are a xed.

This fact has a bearing also on gender motion as de ned in sec. 3.1, which is in itself a type of derivation. Gender motion a fects animate

(mostly human) nouns; it derives feminine nouns from masculine bases

Koch (2001: 1066) shows that gender (or noun class) alternation in Swahili and in Ital-ian functions basically in the same way as a formal motivation device.


18/40

- 209

through the addition of a speci c derivational su x. This su x is for-mally di ferent from the stem building vowel that indicates gender shiftthrough change of in ectional class. In addition, the semantic motivationof a purely derivational motion su x is higher than the possible semanticmotivation of a stem building su x such as the thematic vowel. ConsiderTable 3:

Table 3. Animate beings: gender shift vs. gender motion.

Masculine Femininei vkawolf vk (Skr.)ii devagod dev (Skr.)iii thesgod the (Gr.)iv amicusfriend amica (Lat.) v amicofriend amica (It.) vi ragazzo boy ragazza (It.) vii rexking reg-in-a (Lat.) viii gallus cock gall-in-a (Lat.)ix Freund friend Freund-in (Germ.) x geni-tor parent gene-trix (Lat.) xi bevi-tore drinker bevi-trice (It.)

Items (i) through (iv) involve gender shift through change of in ectionalclass. In the feminine forms, the in ectional class is signaled by a vowel which ultimately goes back to su xes containing a laryngeal (*-iH , *-ih ,*-eh ). These su xes, which had become the markers of speci c in ec-tional classes in the Indo-European languages, were morphologically nolonger on the same plane as other derivational su xes already at thestage of late PIE. The development involved will be discussed at length insec. 6. Nouns in (v) and (vi) show how in ectional class change not onlycontinues existing Latin pairs of masculine vs. feminine, as in (v), but is

This table has a synchronic approach: note that here I am not trying to reconstruct a

su x, but rather a derivational process, that is, gender shift through change of in ectionalclass, which occurred after the reanalysis of former su xes containing laryngeals as stembuilding vowels. For this reason, I use forms which are clearly later, asthe: the fact that,as remarked by an anonymous reviewer, the older form wasthes, used both for the mas-culine and for the feminine, far from being a problem, is a demonstration of increasingproductivity of in ectional class change as a derivational process.


19/40


20/40

- 211

to a di ferent paradigm (Pimenova 2004: 253), that is, derivation throughin ectional class change (see further Dressler/Doleschal 1990).

Let us concentrate on Latin (but the same holds for other languages),and let us compare the lexical pairsamicus/amicaand genitor/genetrix. At rst sight, one could have the impression that the process by whichamica is derived fromamicus is the same by which genetrix is derivedfrom genitor : the bases occur with di ferent su xes, as they can be ana-lyzed asamic-a vs.amic-u-s and gene-trc-s vs. geni-tor . However, this isnot the case, as both-a in amica and -u- (


21/40

212

(see Luraghi 2004 for the details of the diachronic evolution of gender andnominal paradigms from Mycenaean to Modern Greek).

In several Indo-European languages in ectional class change fornouns with human referents had the e fect of functioning as a word for-mation device, as in Latin, and brought about gender shift, thus partlyfunctioning as derivational a xes that indicated gender motion. Thisis not true of all languages, however. Let us consider Modern HighGerman. In this language, nouns that end in-e and do not belong tothe -en declension are by the greatest majority feminine (with just afew exceptions, such as Auge and Ende). However, there is no asso-

ciation of-e stems with feminine gender that can motivate derivation:no nouns that refer to human females can be synchronically derivedfrom nouns that refer to human males just by transposing the latter inthe in ectional class of feminine-estems. Instead, and contrary to what we have just observed for Latin and Italian, this type of derivation is pro-ductively accomplished through the motion su x-in. This state of a fairspartly re ects the situation at the Old High German stage: all- stems werefeminine, as they must have been in Proto-Germanic, since they re ectedthe *-(e)h stems, and some feminines derived through gender shift exist,such asherra lady fromherr lord, but the su x-in was already beingused, and by the Middle High German time it had become the produc-tive way to derive feminine nouns (Doleschal 1992: 301). Apparently,the association of gender and in ectional class was not salient enough tobe used for derivation of referentially motivated feminine.

5. G O H C

5.1. Gender and Evaluative MorphologyEvaluative morphology is frequently found cross-linguistically. It involvesderived lexemes such as diminutives, augmentatives, pejoratives, and soon. In this section I will describe some properties of diminutives in Indo-European and other languages, which make it similar to or connected with gender.

According to Pimenova (2004: 252 fn. 10), derivation through in ectional class changeis seldom used in the case of feminine nouns with human female referents in the ancientGermanic languages.


22/40

- 213

In sec. 3.1, it was remarked that one of the di ferences between derivation through gender shift and derivation through derivational a xeslies in the higher semantic motivation of the latter: groups of nouns dis-tinguished through gender shift can be associated by various semanticfeatures, some of which are quite abstract (referential sex, hyperonyms vs. hyponyms, basic level vs. more speci c categories, etc.); often, nocommon semantic motivation can be detected that singles out a groupof nouns. On the other hand, nouns motivated by a derivational a x dis-play a higher degree of semantic coherence. Somewhat in between is thedegree of motivation displayed by diminutives. Indeed, diminution is a

highly polysemous eld cross-linguistically, as various conceptual asso-ciations are at stake, partly pragmatically motivated. Diminutive a xesconvey a vast array of meanings beside the meaning small, sometimesin con ict with each other and ranging from pejorative, but also meliora-tive, etc. (see Dressler/Merlini Barbaresi 1994). Jurafsky (1996) accountsfor semantic extension of the core meaning small in terms of a radialcategory, that is, a category in which members are related to the center,but not necessarily with one another (see Lako f 1987: 91114).

Besides semantic vagueness, diminutives (and evaluative morphologyin general) are similar to gender in various respects: for example, theyare most typical of nouns, but can also spread to other lexical classesin numerous languages (see Bauer 1997). Accordingly, in various Indo-European languages diminutive su xes can also occur with adjectives (ase.g. in Latin and Sanskrit; see further sec. 10.2).

As derivational a xes, diminutives are non-prototypical as they donot change the lexical class of the base: in this respect, diminutives dis-play a property typical of in ection. In addition, there is evidence from

the Bantu languages for agreement triggering diminutives. Remarkably,evaluative morphology in the Bantu languages displays in ectional fea-tures to varying degrees, even though noun classes identi ed by evalu-ative pre xes are apparently less integrated in the noun class systemthan others. This borderline status of diminutives with respect to othernoun classes also has repercussions for the reconstruction of the nounclass system of Proto-Bantu. Traditionally, this system is reconstructed toinclude classes 12/13 of diminutives, a class which is attested for examplein Kikuyu, in which diminutives trigger agreement on adjectives and pro-nouns (see Bauer 1997). However, there is little evidence for these classesin numerous Bantu languages, some of which provide evidence for derivational diminutives disconnected from the noun class system, as shownby Heine and Kuteva: In many southern Bantu languages, such as Venda,


23/40

214

Tonga-Inhambane, or Herero, there is a diminutive su x typically of theform -ana, which is derived from the Proto-Bantu nominal root *-yana child . . .; for example, Venda-ana diminutive su x. (2002: 66)

In the Indo-European languages, diminutive su xes consistently occurafter other derivational su xes if present. In a sense, they are similar tothe thematic vowel, because they follow more prototypical derivationalsu xes; only the (even less prototypical) thematic vowel can followthem (see also Scalise 1984, Bauer 1997). Finally, diminutives are cross-linguistically extremely productive: in a language with some sort ofdiminutive a xes, a diminutive can be built on virtually any noun. Such

a high degree of productivity is typical of in ectional categories, whilederivation is renownedly much more idiosyncratic, and it constitutesanother feature of diminution as non-prototypical derivation (Bauer 1997).The borderline position of diminutives, located in between derivation andin ection, is another common feature with gender.

The Indo-European languages provide evidence for semantic contactsbetween feminine and diminutive. The su x *-n-, originally formingrelational adjectives, later extended to relational nouns. This motivatesfurther extension to feminine, so Latin gallna indicates the female entitythat pertains to gallus. A similar extension has led the su x-in to becomethe standard gender motion su x in German. Another semantic exten-sion based on the relational meaning of the su x brought it to indicatediminutive in some Romance languages, most notably Italian.

In addition, gender assignment by diminutive su xes in the Indo-European languages is remarkable. Contrary to most other derivationalsu xes used in the formation of nouns, diminutives do not assign gen-der in numerous Indo-European languages. This is another formal fea-

ture that diminutives share with in ection, rather than with derivation:indeed, in ectional categories such as number or case do not change thegender of the base (this is obvious, considering that they do not createnew words).

However, in the languages in which diminutives do assign gender tothe base, they invariably assign neuter, also in cases of nouns with humanreferents: this is true of e.g. German, Modern Greek, Serbo-Croatianamong others. In Italian, which, like the other Romance languages, has

Far from being a peculiarity of the Indo-European languages, semantic contiguity ofdiminutive and feminine is cross-linguistically attested: see Jurafsky (1996: 546) on possiblemotivation.


24/40


25/40

216

which is, among other things, the marker of the feminine gender, can havecollective or singulative meaning, depending on the semantics of the base(see sec. 8.2).

The disappearance of the Latin neuter gender partly resulted in aninterplay between gender and number in Romance. The last two lines ofTable 2 highlight this phenomenon in Italian, in which feminine deriva-tives of masculine nouns can result in collective nouns, which are gram-matically singular like various other non-derived collectives (e.g.la follathe crowd); to a limited extent, they also build collective plurals. Notethat in both cases the base also has a count plural:i frutti the fruits,i

muri the walls. Thus, the morpheme-a partly gained back its ancientcollective meaning.In sec. 2.2, it has been remarked that in ectional categories prototypi-

cally feed syntax, while derivational ones feed the lexicon. In this respect,number as an in ectional category has the function of co-indexing thesubject and the verb in the Indo-European languages. That this function isnot exclusively ful lled by number, but can also be carried out by genderis well known: in many Indo-European languages verb forms involvingparticiples also display gender agreement with the subject. In the Semiticlanguages, verbal pre xes co-index the verb with the subject and agree with it both in gender and in number. In Cushitic, only gender marksagreement, while the syntactic function of number is limited:

Number is not an obligatory category. One can use an underived basic formof the noun that is neutral for number in situations where the speci cationof number is considered irrelevant. There is number agreement in the subject marking on the verb, but for several languages, agreement on the verbis with gender and not with number. (Mous 2012)

6. T S * -( ) : F D S M I C

Since Brugmann (1891) the PIE gender system is reconstructed asinvolving two stages: (a) two-gender system: animate/inanimate, mor-phologically corresponding to masculine/neuter in the later system; (b) three-gender system: masculine/feminine/neuter. The feminine gender involvesthe su x *-(e)h , whose original function was to form abstract nouns.The same su x underwent a semantic extension from abstract to col-lective, and changed into the in ectional ending of the nominative/accusative neuter plural, as remarked as early as Schmidt (1889).


26/40

- 217

Over the last two centuries, scholars have been busy trying to nd asemantic motivation that connects the collective meaning to the feminine.Since the change from a derivational su x to an in ectional ending wasalready accomplished before the Anatolian branch split out, it is thoughtthat this change must have preceded the extension of the su x to thefeminine gender, which did not exist in the Anatolian languages. Such arelative chronology raises various morphological and semantic problems, which I am not going to discuss in detail here (see Clackson 2007: 7, andLuraghi 2009a, 2009b for exhaustive discussion); in addition, it envisages acrucial role for a group of human collective nouns formed with *-(e)h , for

which there is no evidence in the IE languages (see further Luraghi 2011).Recently, it has been suggested in Luraghi (2009a) that such problems canbe avoided by assuming a di ferent chronology, which does not connectthe two changes. This hypothesis is summarized in Table 4:

Table 4. Development of the Su x*-(e)h .

1. derivational su x (non-obligatory)2a. neuter nouns:

in ectional su x (nominative/accusative plural, obligatory).2b.i -- stems:marker of in ectional class (thematic vowel, obligatory)

2b.ii rst class adjectives:marker of in ectional class and feminine gender (obligatory)

Stages (1) and (2) are chronologically ordered, whereas stages (2a) and (2b)represent two separate developments:(2a): a derivational su x turns into an in ectional one, preserving (part of) its

meaning;(2b): a non-obligatory, meaningful su x turns into a thematic vowel, i.e. a purelygrammatical, obligatory item, which is also interpreted as the marker of a nounclass (i.e. of a grammatical gender).(Luraghi 2009a: 56)

Note that such a chronology also has the advantage of eliminating prob-lems such as the presence of the su x only in the nominative/accusativeplural (in the case of neuter nouns) or throughout the paradigm (-- stems),

For reasons of space, I refrain from duplicating here my discussion of works concern-ing the reconstruction of the PIE gender system and setting up such a relative chronology,as for example Tichy (1993) or Ledo-Lemos (2003): the interested reader is referred toLuraghi (2009a, 2009b and 2011).


27/40

218

a problem which would arise if one insists in considering the two changeschronologically connected. Indeed, a derivational su x clearly appears inall in ected forms of a noun. In the case of the neuter plural, the su xhas become an ending, and as such it is mutually exclusive of other in ec-tional endings. How exactly such a change has come about goes beyondthe scope of this paper, but a few remarks are in order. The resulting stateof the neuter plural in ection implies that the extension of the su x tocollective happened at a stage when a complete plural paradigm for theneuter was not available: such a stage has been reconstructed for PIE forexample by Risch (1980). Later, when the plural paradigm started to be

built, the (derivational) collective was reinterpreted as nominative/accu-sative neuter plural, and as such integrated into the paradigm.Leaving aside the change that led the su x to become the ending of the

nominative/accusative neuter plural, let us focus on the second change,by which a derivational su x became the marker of an in ectional class.Indeed, this is not so hard to explain: derivational su xes, as we haveseen above, identify groups of entities. In the Indo-European languages,they are attached to the root and appear before in ectional su xes.Thus, a noun with a derivational su x displays this su x throughoutthe paradigm. Thematic vowels (such as-o- of the thematic declension)have a distribution which is partly similar to that of derivational a xes,because they precede in ectional endings and occur throughout theparadigm. If no derivational su xes occur, thematic vowels take the slotof derivational a xes, immediately following the root. (A derivationalsu x may appear before the thematic vowel; this also happens with*-(e)h, as shown in Table 3 in nouns like gall-n-a, rg-n-a, which sharesthe distribution of the thematic vowel in Latin.) We can reconstruct a

situation in which a certain group of PIE nouns took the su x*-(e)h.This group of nouns developed agreement with adjectives and demonstra-tive/anaphoric pronouns: this is the crucial step for the su x to becomea gender marker, that is, the marker of an obligatory, agreement trig-gering category. Agreement must have risen when the su x*-(e)h was

Note further that the nominative/accusative plural is indeed a count plural also inthe languages in which it takes singular agreement, such as for example attic Greek as itcan occur with numerals. This seems to be at odds with its origin as a collective. However,this is not the case: the neuter plural is completely integrated into a number paradigmin the Indo-European languages, and collective is apparently not a possible value for thegrammatical category of number, as pointed out in Corbett (2000: 1179). One must notconfuse the reconstructed origin with the actual function in the attested languages.


28/40

- 219

still a derivational su x, semantically motivated by sex, that is, beforeit underwent the change that led it to become a thematic vowel, that is,the marker of an in ectional class. In what follows, I will turn to the riseof agreement in connection with di ferent types of agreement triggeringmorphemes and di ferent origins of gender markers. Before doing that,however, I would like to elaborate further on the relevance of gender andagreement outside nominal classi cation.

7. G I : S F G

Following Hocketts classical de nition, [g]enders are classes of nounsre ected in the behavior of associated words (1958: 231). Thus gen-der is de ned by its agreement triggering property. Agreement cross-references items that refer to the same entity: it indicates which wordsbelong to the same constituent, and allows recovering the reference ofanaphoras. In other words, agreement feeds syntax. Thus, a de nition ofgender based on agreement is focused on its syntactic function, and alignsgender with in ectional categories.

Emphasis on agreement has led some scholars to minimize the clas-si catory function of gender and point to the identi cation of referentsin discourse as its main function. Following this approach, Dahl states:[i]t is a mistake to think of gender systems as systems for classifyingthings: to the extent that they do so it is secondary to their function tomake it easier to keep track of links between constituents. (2000: 113). Ina similar vein, Frajzyngier and Shay argue that [t]he reason that genderis assigned to a noun is to enable anaphoric reference to that noun in dis-course and deictic reference to that noun in the environment of speech.(2003: 180).

The identi cation of discourse referents is an important function of gen-der: agreement with anaphoric and deictic pronouns allows speakers andhearers to track participants. Indeed, even classi ers (non-obligatory) aresometimes used anaphorically, as shown in (3) and (4), in which numeralclassi ers function anaphorically in Japanese and Mandarin Chinese:

(3) a. nihiki no neko-wo kau

2 NumCl GEN cat ACC raise (I) am raising two cats. Japanese

b. nihiki wo kau 2 NumCl ACC raise (I) am raising two (small animals).


29/40

220

(4) W mi-le ling-tio xn tnzi. Y-tio hi-de, y-tio hng-de. I buy-Asp two-Cl new blanket one-Cl black-DE one-Cl red-DE

I bought two new blankets. One is black and one is red. Mandarin Chinese

Such usage of classi ers in Mandarin Chinese and Japanese is in accor-dance with Greenbergs (1978) theory regarding the origin of gender mark-ers, to which I now turn.

8. H D A C A ?

8.1. From Classi er to Gender Marker In his seminal paper on the creation of gender markers, Greenberg (1978)indicates classifying demonstratives as crucial to the raise of agreement. According to Greenberg, nominal classi ers may spread to demonstratives:such a development leads to agreement between nouns and demonstra-tives, by which nominal classi ers acquire an anaphoric function similarto numeral classi ers in Japanese and Chinese, as shown in examples (3)and (4).

Evidence for the development of classi ers into gender markers isavailable from various languages, and has accordingly been observed bynumerous scholars: see for example Craig (1986a) on Jacaltec (Maya),Claudi (1985) on Zande (Niger-Congo), Grinevald/Seifart (2004) on Ama-zonian languages, and Dixon (1980: 273, 1982: 171) and Aikhenvald (2000:372) on Australian languages. Such a development involves a grammati-calization process: classi ers become obligatory; in the meantime, theylose semantic motivation, that is, markers of noun classes undergo seman-tic bleaching as is common in grammaticalization. Indeed, semantic motivation is lower in all types of gender systems, including highly elaboratedones such as in the Bantu languages, as opposed to classi er systems (seeamong others Aikhenvald 2000, Grinevald/Seifert 2004).

As for the ultimate origin of noun class/gender markers, while thereis ample evidence for Greenbergs hypothesis that they originate fromgeneric nouns, this is not always the case. In particular, the noun classsystem of the Bantu and in general of the Niger-Congo languages doesnot seem to provide any evidence for this development, as argued in Gri-nevald and Seifart (2004: 256).

It must be mentioned that the Niger-Congo system is, by all evidence, very old:so, while these languages provide illustration for the decay of the noun class system, or


30/40

- 221

8.2. Gender Markers from Derivational Su xes

According to Frajzyngier and Shay, noun classes may re ect old deriva-tional morphemes that once coded semantic characteristics and that areno longer transparent (2003: 179). This is the reconstructed origin of theIE feminine gender; some other parallels exist that I will illustrate in thissection.

The most well known instance of a derivational a x involved in gendermarking is found in Afro-Asiatic. The Afro-Asiatic languages have a typi-cal two-gender system, comprising a masculine and a feminine, in whichmany feminine nouns are marked by the su x *-t which also appearson agreeing categories. The gender system of the Afro-Asiatic languageshas several points in common with the system of the Indo-European lan-guages. In the rst place, for nouns with human referents feminine nounsare derived from masculines, which serve as bases. In the second place,there is an important semantic similarity between the feminine gendersin the two language families: similar to the PIE su x *-h , the su x *-t of Afro-Asiatic also occurs in abstract nouns and collectives. In Arabic,for example, feminine derivation has the e fects summarized in Table 5

(adapted from Luraghi 2006):Table 5. The su x-at in Arabic.

1.arafaknow maerif-at-unwisdom2.dam-untears (collective; broken pl.) dam-at-untear (singulative)3.hayawn-un animal (sing. count) hayawn-t-un animals (collective)4.kalb-un dog kalb-at-un she dog

The di ference between the Afro-Asiatic and the IE system lies in the absenceof the neuter from the former language family. This also implies that, con-trary to PIE, there was no gender in Afro-Asiatic before the creation of thefeminine: unfortunately, however, comparative evidence does not allowus to reconstruct such a stage. The Afro-Asiatic family is much older thanthe IE one, and the sex-based two-gender system is one of the most typi-cal and uncontroversial features of all Afro-Asiatic languages. Evidence for

the renewal of relevant morphology, the origin of noun classes can at the best be recon-structed, because the existing evidence points to the existence of a highly grammatical-ized (i.e. obligatory and agreement triggering) system at all stages: see Grinevald/Seifart(2004: 256).


31/40


32/40


33/40

224

of endings for nominative and accusative in the inanimate gender. Agree-ment was achieved in the case of demonstratives by the existence of twoseparate forms, one for animate (*so) and one for inanimate (*tod ), whichpatterned accordingly.

8.4. Di ferences and Similarities among Gender Systems Dependingon Their Origin

In the preceding section, we have seen that gender systems may arise indi ferent way. In particular, they may originate from the grammaticaliza-tion of classi ers (6.1), or of derivational su xes (6.2), or they can derivefrom special agreement patterns triggered by di ferential marking of corearguments (6.3). In the examples shown above, systems that derive fromclassi ers tend to have a large number of genders. On the other hand, thesystems described in 6.2 and 6.3 have a similar development, in that theyall started out with a small number of genders and built on it.

However, it must also be pointed out that gender systems based onolder systems of classi ers and gender systems that rely on older deriva-tional su xes are also similar, since in both cases agreement triggering

gender markers derive from semantically motivated non-agreement trig-gering markers, which had the function of singling out groups of nouns,either at the lexical level (derivational a xes) or at the syntactic level(classi ers). In addition, both types of systems feature overt gender mark-ers, while gender systems that arise from di ferential case marking do not(in Russian, there are no overt markers of animate and inanimate sub-genders within the masculine gender).

In connection with the di ferences described above, Luraghi 2011 arguesthat gender systems may have di ferent primary functions. Gender sys-tems that originate from the grammaticalization of classi catory deviceshave classi cation as their primary function; on the other hand, gendersystems that derive from special agreement patterns are identi ed byagreement: hence, their primary function is to track discourse referents,that is, to indicate, through agreement, co-referring items, such as nounsand modifying attributes or nouns and anaphoric/deictic pronouns. Thetwo functions of gender coexist in all gender systems, but either one ismaximized, partly depending on the origin of a system. In addition, one

The existence of two di ferent demonstratives constitutes a similarity between thePIE development and the development described by Greenberg, which is based onthe existence of classifying demonstratives.


34/40

- 225

can expect the classi catory function to be more relevant in systems withmany gender distinctions: as there are many genders, many categories areavailable for grouping nouns and their referents.

9. T R A P -I -E

If we now turn to the origin of agreement in PIE, we nd two di ferentpatterns. In the case of the animacy based two gender system, agreementoriginated out of di ferential marking of subjects and objects: as is wellknown, in PIE the neuter did not feature any speci c ending for the nomi-native and the accusative, while the animate gender featured the endings*-s and *-m. This situation is re ected in the athematic declension of theIE languages; later, when the thematic declension was created, the end-ing *-m of the animate accusative was extended to the nominative andthe accusative of neuters. Thus, contrary to animate nouns, inanimate(neuter) ones did not have a distinction between the nominative and theaccusative; in addition, in the athematic declension this lack of distinc-tion was accompanied by the absence of an ending. In such a situation,

agreement did not imply the extension of a speci c marker to pronounsor adjectives: since pronouns and adjectives referred to the same (animateor inanimate) entities, their in ection followed the same pattern of cor-responding nouns.

The situation is completely di ferent for the creation of agreement inthe sex based three gender system of late PIE. Contrary to the masculine(which continues the animate gender) and the neuter, the feminine bearsan overt gender marker, which was a derivational su x originally a xedto nouns. So the rise of agreement implies extension of the marker toagreeing categories.

The di ferent origin of agreement classes, and hence of gender distinc-tions, in PIE also has implications for possible parallels between genderand other word formation devices: as noted above, the animate and theinanimate (neuter) do not feature overt gender markers, and do not dis-play di ferent in ectional classes; in fact, they are identical except for thenominative/accusative. Indeed, there is no systematic motion relationbetween groups of animate and neuter nouns similar to those in Table 3

which involve masculine and feminine.It is commonly thought that the spread of the su x *-h rst targeteddemonstrative pronouns. This is in accordance with Greenbergs (1978)theory regarding the acquisition of gender markers, and with develop-ments observed in various non-Indo-European languages in which gender


35/40

226

systems came about as the result of the grammaticalization of earlier sys-tems of classi ers, as described for example in Claudi (1985) in the case ofZande. In addition, also in gender systems in which a new agreement classis the outcome of some special patterns of case marking, there is evidencefor pronouns to be the rst targets of agreement: this has been observedfor example in Old Russian (see Huntley 1980).

In the next section, I would like to explore another possibility, that is,that agreement spreads rst to modifying adjectives.

10. A D M

10.1. Modi ers from Agent Nouns

As remarked in sec. 4.1, agent nouns can build deverbal adjectives in Latin.Indeed, this is true of other languages, as we will see in this section. Basedon data from Romance, Russian and Modern Greek, Dressler and Dole-schal (1990) describe what they de ne agreement via derivational mor-phology. Consider the following examples:

(5) licuit victricem classem in Africam traicere atque intra paucos dies sine ullocertamine Carthaginem delere

We might have carried over our victorious eet to Africa, and in a few dayshave destroyed Carthage without any opposition. Liv. 21,41

(6) Haec inquit a me Vercingetorix bene cia habetis quem proditionis insimulatiscuius opera sine uestro sanguine tantumexercitum uictorem fame paene con-sumptum uidetis. These bene tssays Vercingetorixyou receive from me, whom you accuseof treasonme, by whose exertions you see so powerful and victorious an

army almost destroyed by famine, without shedding one drop of your blood.Caes. Gal. 7.20.12

In (5) and (6) the formsuictricem and uictorem agree with their headnoun in gender through the su x: as the third declension, to which bothuictrixand uictor belong, is not connected with gender, the two su xesacquire the status of gender markers for this type of deverbal adjectives.This phenomenon is widespread in Italian:

(7) uno sguardo rivelatore/ una risposta rivelatrice a revealing look (m.) / a revealing answer (f.)

(8) *sguardo rivelatrice / *risposta rivelatore


36/40

- 227

In Italian, adjectives display two patterns of in ection: (a) masculine in-o/-i , feminine in-a/-e; (b)-e/-i with no gender distinction (this continuesthe rst and second class adjectives of Latin). In the case of adjectives dis-playing the second pattern, those with the su x-tore/-trice constitute asub-class, marked for gender by the derivational su x (Ricca 2004). Sucha phenomenon is typical for deverbal adjectives formed with agent suf-xes with gender speci c forms, as for example Modern Greek:

(9) I niktria omda the victorious team /O nikitrs Olimbiaks the victoriousOlimbiakos

(Christo dou et al. 19901991: 72)

Example (8) shows that agreement via the su x is obligatory if a deverbaladjective of this type occurs. However, similarities between this remark-able type of agreement and gender agreement as it was brought about bythe spread of the su x *-h in PIE ends here: deverbal adjectives describedin this section do by no means occur only or especially with agent nouns which bear the same su x. Indeed, the su x has not spread historicallyfrom agent nouns to accompanying adjectives: rather, agent nouns have

come to be used as adjectives. In addition, there is no evidence for any occur-rence of the su x on pronouns, or for its possible function as anaphora.

10.2. Diminutives

As remarked in sec. 5.1, diminutive su xes can also occur with adjec-tives in several Indo-European languages. To strengthen the meaning ofthe diminutive, a diminutive su x can then appear both on a noun andon the adjectives modifying it: (10) is a well known example from Latin, which shows not only diminutive adjectives, but agreement in the cat-egory of diminution between the head noun and the modi ers:

(10) Animula vagula blandula (Hadrian)

Examples can be found from other languages. Below I give some fromItalian, in order to highlight the di ference with the data examined in thepreceding section. Note that similar examples could be mentioned with

augmentatives:(11)Spero che prenda un votino un pochino discretino (Nicola Grandi, p.c.) I hope s/hell get a slightly(dim.) decent(dim.) note(dim.)


37/40


38/40

- 229

the lexicon, as a means for creating new words and motivating groups of words. I ended indicating some possible evidence for mechanisms thatmay lead to the rise of agreement through the spread of a derivationala x.

R

Aikhenvald, Alexandra. 2000.Classi ers. A Typology of Noun Categorization Devices. Oxford:Oxford University Press.

. 2004. Gender. In G.E. Boij, Ch. Lehmann, J. Mugdan and S. Skopetas (eds.), Morfolo- gie/Morphology. Ein Handbuch zur Flexion und Wortbildung/A Handbook on In lectionand Word Formation, 2. Halbband, 103145. Berlin and New York: Mouton de Gruyter.

. 2006. Classi ers and noun classes, semantics. Volume 1 (article 1111) of Encyclope-dia of Language and Linguistics, 2nd edition, edited by Keith Brown, 46370. Oxford:Elsevier.

Bauer, Laurie. 1997. Evaluative morphology in search of universals.Studies in Language 21/3.53375.

Booij, Geert. 1996. Inherent versus contextual in ection and the split morphology hypoth-esis. In G. Booij and J. van Marie (eds.),Yearbook of Morphology 1995 , 116. Dordrecht:Kluwer.

Brugmann, Karl. 1891. Zur Frage der Entstehung des grammatischen Geschlechtes. Beitrge zur Geschichte der deutschen Sprache und Literatur ( PBB ) 15.52331.

Christo dou, Anastasia, Ursula Doleschal and Wolfgang U. Dressler. 19901991. Genderagreement via derivational morphology in Greek.Glossologa 910.6979.Clackson, James. 2007. Indo-European Linguistics. An Introduction. Cambridge: Cambridge

University Press.Claudi, Ulrike. 1985. Zur Entstehung von Genussystemen. berlegungen zu einigen theore-

tischen Aspekten, verbunden mit einer Fallstudie des Zande. Hamburg: Buske.Corbett, Greville. 1991.Gender . Cambridge: Cambridge University Press.. 2000. Number . Cambridge: Cambridge University Press.Craig, Colette (ed.). 1986. Noun Classes and Categorization. Amsterdam and Philadelphia:

Benjamins.. 1986a. Jacaltec Noun Classi ers: A Study in Language and Culture. In Craig (1986),

26393.Dahl, sten. 2000. Animacy and the notion of semantic gender. In B. Unterbeck,

M. Rissanen, T. Nevalainen and M. Saari (eds.),Gender in Grammar and Cognition: I. Approaches to Gender, II. Manifestations of Gender , 99115. Berlin and New York:Mouton de Gruyter.

Dixon, R.M.W. 1980.The Languages of Australia. Cambridge: Cambridge University Press.. 1982.Where Have All the Adjectives Gone? and other essays in Semantics and Syntax.

Berlin, New York and Amsterdam: Mouton Publishers.Doleschal, Ursula. 1992. Movierung im Deutschen. Eine Darstellung der Bildung und Ver- wendung weiblicher Personenbezeichnungen. Munchen: Lincom Europa.

Dressler, Wolfgang U. 1989. Prototypical Di ferences between In ection and Derivation. Zeitschrift fr Phonetik, Sprachwissenschaft und Kommunikationsforschung42.310.

Dressler, Wolfgang U., and Ursula Doleschal. 1990. Gender Agreement via DerivationalMorphology. Acta Linguistica Hungarica40.11537.Dressler, Wolfgang U., and Lavinia Merlini Barbaresi. 1994. Morphopragmatics: Diminutives

and intensi ers in Italian, German and other languages. Berlin: Mouton de Gruyter.Dressler, Wolfgang U., Willi Mayerthaler, Oswald Panagl and Wolfgang U. Wurzel. 1987. Leitmotifs in Natural Morphology. Amsterdam and Philadelphia: Benjamins.


39/40

230

Frajzyngier, Zygmunt. 1997. Grammaticalization of number: From demonstratives to nomi-nal and verbal plural. Linguistic Typology1.193242.

. 2008. A Grammar of Gidar . Frankfurt a. M.: Lang.Frajzyngier, Zygmunt, and Erin Shay. 2003. Explaining Language Structure through System Interaction, Amsterdam and Philadelphia: Benjamins.

Grandi, Nicola. 2002. Morfologie in contatto. Le costruzioni valutative nelle lingue del Medi-terraneo. Milano: Franco Angeli.

Greenberg, Joseph H. 1978. How does a language acquire gender markers? In J. Greenberg(ed.), Universals of human language. Vol. 3:Word structure, 4782. Stanford: StanfordUniversity Press.

Grinevald, Colette, and Frank Seifart. 2004. Noun classes in African and Amazonian lan-guages: towards a comparison. Linguistic Typology 8/2.24385.

Heine, Bernd, and Tania Kuteva. 2002.World lexicon of grammaticalization. Cambridge:Cambridge University Press.

Hockett, Charles F. 1958. A course in modern linguistics. New York: Macmillan.Huntley, David. 1980. The evolution of genitive-accusative animate and personal nouns inSlavic dialects. In: J. Fisiak (ed.), Historical Morphology, 189212. The Hague: Mouton.

Jurafsky, Dan. 1996. Universal Tendencies in the Semantics of the Diminutive. Language 72.53378.

Kihm, Alain. 2005. Noun class, gender, and the lexicon-syntax-morphology interfaces: Acomparative study of Niger-Congo and Romance languages. In G. Cinque and R.S. Kayne(eds.),The Oxford Handbook of Comparative Syntax, 459512. Oxford: Oxford UniversityPress.

Koch, Peter. 2001. Lexical typology from a cognitive and linguistic point of view. InM. Haspelmath, E. Knig, W. Oesterreicher and W. Raible (eds.), Language Typologyand Language Universals. An International Handbook. Vol. 2., 114278. Berlin: Walter deGruyter.

Krishnamurti, Bhadriraju. 2001.Comparative Dravidian linguistics: current perspectives.Oxford: Oxford University Press.

. 2003.The Dravidian Languages. Cambridge: Cambridge University Press.Kuryowicz, Jerzy. 1964.The In lectional Categories of Indo-European.Heidelberg: Carl

Winter.Lako f, George. 1987.Women, re, and dangerous things: what categories reveal about the

mind . Chicago: University of Chicago Press.Ledo-Lemos, Francisco. 2003. Femininum Genus. A Study of the Origins of the Indo-European Feminine Grammatical Gender . Mnchen/Newcastle: LINCOM Europa.

Luraghi, Silvia. 2004. The evolution of the Greek nominal paradigms: economy and case

syncretism from Mycenean to Modern Greek.Classica et Mediaevalia 55.36179.. 2006. La nascita del genere femminile in indoeuropeo. In Luraghi/Olita (2006),89106.

. 2009a. The origin of the feminine gender in Indo-European. An old problem in anew perspective. In V. Bubenik, J. Hewson and S. Rose (eds.),Grammatical Change in

Indo-European Languages, 313. Amsterdam and Philadelphia: Benjamins.. 2009b. Indo-European nominal classi cation: From abstract to feminine. In S.W.

Jamison, H. C. Melchert and B. Vine (eds.), Proceedings of the 20th Annual UCLA Indo- European Conference, 11531. Bremen: Hempen.

. 2011. The origin of the Proto-Indo-European gender system: Typological consider-ations. Folia Linguistica45/2, 43564.

Luraghi, Silvia, and Anna Olita (eds.). 2006. Linguaggio e genere. Roma: Carocci.. 2006a. Introduzione. In Luraghi/Olita (2006), 1527.Malzahn, Melanie. 2000. Die Genese des indogermanischen Duals. In M. O tsch and Ch.

Zinko (eds.),125 Jahre Indogermanistik in Graz, 291315. Graz: Leykam.


40/40

- 231

Meillet, Antoine. 1921. La catgorie du genre et les conceptions indo-europennes. In A. Meillet (ed.), Linguistique Historique et Linguistique Gnrale, 21129. Paris:Champion.

Mous, Maarten. 2012. Cushitic. In Z. Frajzyngier and E. Shay (eds.), Afro-Asiatic languages,342422. Cambridge: Cambridge University Press.

Mufwene, Salikoko. 1980. Bantu noun class pre xes: In ectional or derivational? In J. Krei-man and O.J. Ojeda (eds.),CLS Working Papers in Linguistics, 24658. Chicago.

Pimenova, Natalia. 2004. Nominale Stammbildungssu xe als Derivationsmittel im(Gemein)germanischen. In J. Clackson and B.A. Olsen (eds.), Indo-European Word For-mation, 24868. Copenhagen: Museum Tusculanum.

Ricca, Davide. 2004. Cumulative exponence involving derivation. In Morphology and itsdemarcations, 2078. Amsterdam and Philadelphia: Benjamins.

Risch, Ernst. 1980. Betrachtungen zur indogermanischen Nominal exion. In Ch. Lehmannand G. Brettschneider (eds.),Wege zur Universalien forschung, 25967. Tbingen: Narr.

Scalise, Sergio. 1984.Generative Morphology. Dordrecht: Foris.Schmidt, Johannes. 1889. Die Pluralbildungen der indogermanischen Neutra. Weimar:Bhlau.

Seifart, Frank. 2009. Multidimensional typology and Miraa class markers. In P. Epps and A. Arkhipov (eds.), New Challenges in Typology. Transcending the Borders and Re ningthe Distinctions, 36585. Berlin and New York: Mouton de Gruyter.

Thornton A.M. 2003. Lassegnazione del genere in italiano. In F. Snchez Miret (ed.), Actasdel XXIII Congreso Internacional de Lingstica y Filologa Romnica, vol. I, 46781. Nie-meyer, Tbingen.

Tichy, Eva. 1993. Kollektiva, Genus femininum und relative Chronologie im Indo-germanischen. Historische Sprachforschung 106.119.

Walter, Heribert. 1982. Genus- und Nominalklassensysteme und die Dimension der Appre-hension. In H. Seiler and Ch. Lehmann (eds.) Apprehension. Das sprachliche Erfassen vonGegenstnden. Teil I: Bereich und Ordnung der Phnomene, 21728. Tbingen: Narr.

Zubin, David and Klaus M. Kpke. 1986. Gender and folk taxonomy. In Craig (1986),13980.

Documents

Luraghi Gender and Word Formation