Gender, Grammatical 749

Page R (2003). ‘‘‘Cherie: lawyer, wife, mum’’: contradictory Shaw S (2000). ‘Language, gender and floor apportion-
patterns of representation in media reports of Cherie ment in political debates.’ Discourse and Society 11(3),
Booth/Blair.’ Discourse and Society 14(5), 559–580. 401–418.
Peterson S & Sisson Runyan A (1993). Global gender Sreberny-Mohamadi A & Ross K (1996). ‘Women MPs and
issues. Boulder/San Francisco/Oxford: Westview Press. the media: representing the body politic.’ In Lovenduski J
Puwar N (1997). ‘Gender and political elites: women in the & Norris P (eds.) Women in politics. Oxford: Oxford
House of Commons.’ Sociology Review 7(2), 2–6. University Press. 105–117.
Romaniuk T (2004). ‘‘Looks like a mayor, thinks like a Tannen D (1996). Talking from 9 to 5. Women and men at
mayor, talks like a neighbour’’: the discursive con- work: language, sex, and power. London: Virago.
struction of David Miller. M.A. Thesis: York University, Walsh C (2001). Gender and discourse: women in politics,
Toronto, Canada. the church, and the environmental movement. New
Ross K & Sreby-Mohammadi A (1997). ‘Playing house – York: Longman.
gender, politics and news media in Britain.’ Media, Wodak R (2003). ‘Multiple identities: the roles of females
Cultura and Society 19(1), 101–109. parliamentarians in the EU Parliament.’ In Holmes J &
Sedgemore B (1995). The insider’s guide to parliament. Meyerhoff M (eds.) The handbook of language and
Cambridge: Icon Books Limited. gender. Oxford: Blackwell.

Gender, Grammatical
G G Corbett, University of Surrey, Surrey, UK There are many kinds of noun: those with four sylla-
ß 2006 Elsevier Ltd. All rights reserved. bles, those denoting agents, and so on. But ‘gender’ is
normally reserved for kinds or classes of noun
that are, as Hockett (1958: 231) put it, ‘‘reflected in
Gender is a feature of special interest because it the behavior of associated words.’’ To make valid
provides a dramatic demonstration of just how differ- comparisons, it is important to have this externally
ent languages can be. For many of the Indo-European motivated classification. The relevant reflection in the
languages it is an important feature that is realized in a behavior of associated words is ‘agreement’ (includ-
high proportion of utterances. In most Daghestanian ing for some linguists antecedent-anaphor relations).
languages, such as Tsakhur, it is still more salient: The noun inventory is divided into different kinds, or
it suffuses the syntax and morphology, appearing on genders, according to the different agreements they
some unlikely agreement targets. In almost all the take. When this is done, we find that in the more
languages of the Austronesian family, it is simply familiar languages, the different kinds or genders
missing. This article first examines the definitions of have a semantic core based on sex (thus, Russian
gender to ensure that we are comparing like with like. nouns divide into three kinds; nouns denoting males,
There has been careful research to do this, and it though not only these, group together, and those
will become clear that within the languages that do denoting females also group in another gender). In
have gender there is a considerable variety of possible other languages the structures may be very similar,
gender systems. This is particularly apparent in the but the semantic core may not based not on sex;
ways in which speakers assign nouns to genders. Gen- for instance, it may be based on human versus non-
der systems may have sex as a component, as human or animate versus inanimate. Thus a language
in languages with masculine and feminine genders; has a gender system only if noun phrases headed by
but, equally, sex may be irrelevant, as in the Algon- nouns of different types control different agreements.
quian languages, where the distinction is between No amount of marking on a noun can prove that it
animate and inanimate. This article outlines the has gender; the evidence that nouns have gender in a
distribution of gender in the world’s languages, given language lies in the agreement targets that
and finally considers prospects for investigating the show gender. An illustration of this from the Nakh-
feature further. Daghestanian language Bagvalal (Kibrik, 2001:
64–65) will show this. Each noun in these examples
is representative of many more:
Terms
(1) w-es̄ˇa-w waša
The word ‘gender’ derives from Latin genus via Old MASC.SING-plump-MASC.SING boy
French gendre, originally meaning ‘kind’ or ‘sort’. ‘a plump boy’
750 Gender, Grammatical

forms are syntactically appropriate, nouns like otec

(2) j-es̄ˇa-j jaš
FEM.SING-plump-FEM.SING girl ‘father’ and stol ‘table’ take the possessive adjective
‘a plump girl’ moj ‘my’ (moj otec ‘my father’). Since this gender
includes many nouns denoting males, it is convention-
(3) b-es̄ˇa-b ¿ama ally labeled ‘masculine,’ even though the majority of
NEUT.SING-plump-NEUT.SING donkey
nouns it includes, like stol, do not denote males.
‘a plump donkey’
A second agreement class consists of nouns like mat’
Attributive modifiers show that the nouns of Bag- ‘mother’ and kniga ‘book’, which take moja, and
valal divide into three agreement classes. In the plu- make up the feminine gender. Finally, there are nouns
ral, examples like (1) and (2) fall together, and so the like pis’mo ‘letter’, which take moe, and comprise the
plural does not yield additional agreement classes. No neuter gender.
more does changing the case. Let us try changing the For many languages there is no dispute as to the
type of target, by looking at a verbal predicate: number of genders, but there are others where the
question is complex. For instance, Romanian gender
(4) waša w-iRi
boy MASC.SING-stop
has provoked a lengthy debate. The analytical prob-
‘the boy stopped’ lem of determining the number of genders and the
tests for deciding the gender of a given noun depend
(5) jaš j-iRi on separating out the sets into which nouns are
girl FEM.SING-stop
divided (the ‘controller genders’) from the number
‘the girl stopped’
of different genders marked on agreement targets
(6) ¿ama b-iRi (the ‘target genders’). Frequently the two match up,
donkey NEUT.SING-stop as in French or German, but in several languages they
‘the donkey stopped’ do not. A full treatment of the subject is not possible
Although the morphological realization is some- here; for a detailed account and extensive references
what different (prefixal only instead of being prefixal see Corbett (1991).
and suffixal), the verbal predicate does not provide
further agreement classes. Bagvalal has three agree-
Gender Assignment
ment classes, and each should be recognized as a
gender. Given that there are the analytic means to establish
For languages in the Nakh-Daghestanian family, the genders of a language and to determine which
like Bagvalal, the term used is often ‘noun classes.’ nouns belong in which gender, we now approach the
But there is no substantive difference between ‘gen- question from the opposite perspective, and ask how
ders’ and ‘noun classes’; the different terms arise in nouns are allotted to the genders of a given language.
different linguistic traditions. For instance, Kannada In other words, how does a native speaker ‘know’
(a Dravidian language) has three genders, to which what gender a noun belongs to? Models of the
nouns are assigned by rules similar to those of Bagva- mechanisms by which nouns are allotted to genders
lal. By tradition, Kannada is said to have three gen- are called ‘assignment systems.’ Assignment may in-
ders, and Bagvalal three noun classes. We shall use the volve two sorts of information about the noun: its
term ‘gender’ for both. meaning and its form. To illustrate these, we will
take some well-established clear examples. There
The Number of Genders are others that involve more complex combinations
of systems.
The approach to gender sketched in the first section
of this article is based on the notion of ‘agreement
Strict Semantic Assignment Systems
class’ (Zaliznjak, 1964). For two nouns to be in the
same agreement class, they must take the same agree- A strict semantic assignment system is found in
ments under all conditions – that is, if we hold con- Bagvalal (Kibrik, 2001: 64–66), as illustrated in the
stant other features such as case and number. (The earlier examples. Nouns denoting male humans (and
agreements and their domains can be extremely only those) are masculine, while those denoting fe-
varied; Corbett, forthcoming.) If two nouns differ in male humans are feminine. All remaining nouns are
their agreements when factors such as case and num- neuter. Thus, waša ‘boy’ is masculine, jaš ‘girl’ is
ber are held constant, then they belong to two differ- feminine, and ama ‘donkey’ is neuter. The meaning
ent agreement classes and normally they will belong is sufficient, and no information about the form of a
to two different genders. To take another example, in noun is needed to determine its gender. Kala Lagaw
Russian, in environments where nominative singular Ya, spoken on the Western Torres Straits Islands, has
 Gender, Grammatical 751

a two-gender system, also with strict semantic assign- genders, primarily for male humans and non-human
ment: nouns denoting males (and the moon, to which animates (I), female humans (II), non-flesh food (III),
we return below) are masculine, and all others belong and the residue (IV). There are various nouns for
in the feminine gender (Bani, 1987). In Diyari (Dieri), which additional principles are required. For exam-
a language of South Australia, we find the converse: ple, the moon is in the first (masculine) gender and the
there is a gender for nouns with female referents (such sun is in the second (feminine) gender. This recalls
as women, girls, doe kangaroos), and the other Kala Lagaw Ya. In Dyirbal mythology, as indeed in
gender is for all remaining nouns (Austin, 1981). much of Australia, the moon is the husband of the
Strict semantic systems are found in various parts sun. For such instances, the mythological significance
of the world and are particularly prevalent in the of referents can determine gender assignment. Dixon’s
Dravidian family, where there are three-gender sys- data and analysis continue to provoke debate, and
tems (as in Kannada and Tamil) and two-gender there is similar debate about the role of worldview
systems (as in the Parji-Gadaba languages). in the gender assignment system of Ojibwa (and of
other Algonquian languages of North America).
Predominantly Semantic Assignment Systems In all the languages discussed so far, the meaning
of the noun determines gender. In the strict as-
Many languages have semantic assignment rules signment systems, the rules are transparent; in the
that are, however, less comprehensive than those of predominantly semantic systems, there are excep-
Bagvalal. Predominantly semantic assignment is tions, although in some cases these may be explicable
found in Tsakhur, which like Bagvalal is a Nakh- once the cultural setting is taken into account. If we
Daghestanian language. Its four genders are shown ask which are the semantic criteria on which semantic
in Table 1. systems can be based, we see recurring patterns.
Assignment to genders I and II is straightforward: Quite often we find animate/inanimate, human/non-
I is for male humans (also gods, angels, and so on), human, and male/female. Sometimes there is a gender
while II is for female humans (and female mythical for diminutives, as in various Bantu languages. There
beings). Most remaining animates are assigned to gen- are also less usual genders, such as that for non–flesh
der III, although a few belong to gender IV, along with food (Dyirbal) and the gender for insects (found in the
some mythical beings. Inanimates are found in gen- Rikvani dialect of the Nakh-Daghestanian language
ders III and IV, and it is hard to discern a pattern Andi). A criterion that is sufficient to define a gender
(Mel’nikov and Kurbanov, 1964; Ibragimov, 1990: in one language may be just one factor in the assign-
54–56 and references there; Kibrik, 1999: 48–49). ment of gender in another. Thus the Bantu language
Languages of this general type are widespread. For Chichewa (a dialect of Nyanja) has a gender for
some of them, researchers have proposed that abstract diminutives, while in the Omotic language Dizi,
semantic criteria partly miss the point, and that if we diminutives together with nouns denoting females
can gain a better grasp of the worldview of the speak- form a gender.
ers, we can then understand the assignment system
more fully. The most discussed case is that of Dyirbal,
Formal Assignment Systems
North Queensland (Dixon, 1972), which has four
In many languages, the semantic rules assign many
nouns to the appropriate gender, but they also simply
Table 1 Gender assignment in Tsakhur fail to apply to many others. In Russian and in many
Criterion Gender Examples Glosses other Indo-European languages, for sex-differentiables
(those where sex is salient or of importance to
Male I (masculine) baba grandfather humans), nouns denoting males are masculine and
rational dakj father
Female II (feminine) jedj mother
those denoting female are feminine. But these rules
rational jiš daughter have nothing to say about the majority of the noun
Animate III (animate) aImale donkey inventory. While in languages like Bagvalal, the nouns
(also some not assigned by the semantic rules (the ‘remainder’ or
inanimates) ‘semantic residue’) all belong to a single gender, in
balkan horse
(dama) (river)
many languages they are found in more than one
(lat) (trough) gender, even in all the genders (as in Russian). Here
Other (including a IV (neuter) kalle head there are additional rules for assigning nouns to gen-
few animates) ders, but according to their form. Languages may use
sen year semantic rules, or semantic and formal rules, but not
(kabaj ) (butterfly)
only formal assignment rules. In no language are
752 Gender, Grammatical

nouns assigned to genders, as defined earlier, by pure- Table 2 Russian nouns belonging to the semantic residue
ly formal rules. An example would be a language in Masculine Feminine Neuter
which there were two agreement classes, and the
nouns in the first all ended in a consonant cluster, dub ‘oak’ sosna ‘pine’ derevo ‘tree’
and those in the second did not, and there was no stvol ‘(tree) trunk’ doska ‘plank’ brevno ‘log’
čaj ‘tea’ voda ‘water’ moloko ‘milk’
semantic regularity for the distribution of nouns. ogon’ ‘fire’ peč’ ‘stove’ plamja ‘flame’
I claim that this hypothetical type does not exist. okean ‘ocean’ reka ‘river’ more ‘sea’
Formal assignment rules may appeal to two types of avtomobil’ ‘car’ mašina ‘car’ taksi ‘taxi’
information: phonological and morphological. den’ ‘day’ noč ’ ‘night’ utro ‘morning’
čas ‘hour’ minuta ‘minute’ vremja ‘time’
nerv ‘nerve’ kost’ ‘bone’ serdce ‘heart’
Phonological The clearest example of assignment glaz ‘eye’ brov’ ‘eyebrow’ yeko ‘eyelid’
depending on phonological information yet found is lokot’ ‘elbow’ lodyžka ‘ankle’ zapjast’e ‘wrist’
provided by Qafar (Afar), an East Cushitic language flag ‘flag’ èmblema ‘emblem’ znamja ‘banner’
spoken in northeastern Ethiopia and in Djibouti (data
Nouns in bold are inflected in Table 3.
from Parker and Hayward, 1985). Qafar has rather
standard semantic assignment rules, namely, that for
sex-differentiable nouns, those denoting males are
masculine and those denoting females are feminine. sex-differentiables, nouns denoting males are mas-
It is the nouns that fall outside these rules – the culine and those denoting females are feminine. But
residue – that are of interest. For them the phonologi- unlike the situation in languages like Bagvalal, the
cal rules apply: nouns whose citation form ends in an nouns not covered by these rules – the semantic resi-
accented vowel are feminine (for example, catò due – are not simply assigned to the neuter gender.
‘help’), while all others are masculine (for example, Rather, in Russian, the residue is shared between the
gilàl ‘winter’, which does not end in a vowel, and three genders, with the neuter gender not even receiv-
tàmu ‘taste’, which ends in an unaccented vowel). ing the majority. It seems unlikely that we are failing
These rules operate with few exceptions. There is an to spot additional semantic criteria; see (Table 2).
interesting twist here: nouns denoting males and If the morphology of the nouns is examined in-
females typically fit with the phonological assignment stead, then progress can be made. There are arguably
rules, too (for example, bàqla ‘husband’ and barrà four main inflectional classes in Russian, each with
‘woman, wife’). Given the data presented so far, we several thousands of nouns. There are six cases and
might try to dispense with semantic rules for Qafar two numbers (although no paradigm has 12 dis-
and treat it as being of a quite different type. Howev- tinct forms because of various syncretisms). Table 3
er, there are crucial examples like abbà ‘father’, which provides the singular forms.
is masculine, even though it ends in an accented Of course, a speaker needs to know the inflectional
vowel. Similarly, gabbixeèra ‘slender-waisted female’ behavior of a noun. On the basis of that information,
is feminine, even though the accent is not final. Qafar the assignment rules are straightforward. Nouns in
has straightforward phonological assignment rules. inflection class I are masculine, those in classes II and
When the semantic and phonological rules are both III are feminine, and those in IV are neuter. As with
applicable, they almost always give the same result in Qafar, we might wonder whether the semantic assign-
Qafar. However, in cases of conflict, the semantic ment rules are superfluous, since otec ‘father’ is in
rules take precedence, as is the normal situation in class I, while sestra ‘sister’ is in class II, and
gender assignment systems. There are many more mat’ ‘mother’ is in class III, and therefore many sex-
example of phonological assignment, although few differentiable nouns would be assigned correctly by
if any are as clear as those of Qafar. At the other the morphological assignment rules. But there are
extreme we find French, often claimed to have no also instances where this is not so. Thus, deduška
system to its gender assignment. Yet French has a ‘grandfather’ denotes a male but is in class II, whose
phonological assignment system. For example, of nouns are typically feminine; it is masculine. Nouns
938 nouns ending in /ED/, 99 percent are masculine like this show, once again, that we do not find lan-
(le pain [pED] ‘the bread’) (for details see Tucker et al., guages where formal assignment rules are sufficient.
1977; Hardison, 1992). Further rules are required in Russian for indeclinable
nouns (like taksi ‘taxi’, which is indeclinable and neu-
Morphological The morphological assignment sys- ter). Morphological assignment systems are found
tem that has received the most attention is probably in various other Indo-European languages. Looking
that of Russian. Earlier it was noted that in Russian, further afield, in Arabic, too, gender is assignable in
as in many other Indo-European languages, for the main according to morphology (Cowell, 1964:
 Gender, Grammatical 753

Table 3 Inflectional classes in Russian Prospects

I II III IV Studies on grammatical gender continue to appear
Nominative dub sosna noč’ brevno regularly. There are four areas in which we can hope
Accusative dub sosnu noč’ brevno for progress.
Genitive duba sosny noči brevna
Dative dubu sosne noči brevnu Data
Instrumental dubom sosnoj noč’ju brevnom
Locative dube sosne noči brevne Studies of previously unresearched or little-researched
Gloss ‘oak’ ‘pine’ ‘night’ ‘log’ languages continue to extend our knowledge of
gender systems. For example, in Ngan ’gityemerri a
Daly language spoken in northern Australia) there is a
372–375). In Kuot, a language isolate of New Ireland, development from generic classifiers into genders
Lindström (2002: 147–164, 176–194) suggested that (Reid, 1997); a comparable situation, but with a very
there were 11 inflectional classes for nouns, and the large system, is found in Miraña (Boa), a Witotoan
assignment of nouns to the two genders could be language spoken in Colombia, as described by Seifart
predicted to a significant extent on the basis of the (2004). It is possible to have gender and classifier
inflectional class (in addition to a semantic rule based systems co-existing, as in Tariana (Aikhenvald,
on sex). 2003). Among the better known languages, some
The clearest examples have been chosen here for studies of gender are based on a few convenient
illustration. In some languages, the different types of parts, while fuller studies of the whole morphosyn-
assignment criteria overlap and interact in complex tax, to the extent that it is relevant to gender, and of a
ways. Sometimes there are small clusters of nouns substantial portion of the lexicon, can contribute sig-
subject to semantic criteria (apart from those covered nificantly to our understanding of gender. The incor-
by the main semantic rules), even within systems poration of loanwords can be seen as a continuously
where formal rules have a major role (as shown for running experiment on an assignment system, offer-
German by Zubin and Köpcke, 1986; Köpcke and ing the possibilities for statistical analysis, as in
Zubin, 1996). Even in these complex cases, whenever Kilarski’s (2003) account of English borrowings into
gender languages are analyzed in sufficient detail, the Danish, Norwegian, and Swedish.
gender of the vast majority of nouns turns out to be
predictable. Lexicology

Gender is a key feature for lexicology, since it is

Distribution crucial for understanding the nature of lexical entries.
In a gendered language, the gender of a noun must be
Given this typology of gender systems, we can now
available. Yet as we have seen, it does not normally
ask how they are distributed over the world’s lan-
need to be specified, since the information can be
guages. In a sample of 256 languages (Corbett,
derived, by assignment rules, from other information
2005), somewhat over half (144) were found to
that must be stored. We still do not know the full
have no gender system. A gender system with at
range of information that may be relevant (for in-
least two genders (and two-gender systems are com-
stance, the possible semantic features that may deter-
mon) existed in 50 examples in this sample. Three
mine gender assignment). There are further surprises:
genders was about half as common (26 examples),
it is possible for certain nouns to be of different
and four genders was about half as common again (12
genders according to the sex of the speaker; this is
examples). Larger systems, with five or more genders,
found in Garı́funa, a member of the Arawak language
represented a substantial minority (24 languages in
family spoken in Belize, Honduras, Nicaragua, and
the sample). Fula (a Niger-Congo language) has
Guatemala (Taylor, 1977: 60; Munro, 1998). These
around 20 genders, depending on the dialect. Of the
two points raise interesting questions as to the shape
languages with a gender system, the majority had an
of lexical entries.
assignment system based on sex (84 examples), but
28 languages of the sample, notably in the Niger-
Psycholinguistics
Congo and Algonquian families, had systems based
on animacy. And as for the type of assignment system, The issues specified for lexicologists are also increas-
strict semantic and predominantly assignment sys- ingly of concern to psycholinguists (see, for instance,
tems together were found in just under half the lan- van Berkum, 1996). Gender provides a feature that is
guages (53 examples), while a slight majority (59 moderately well described, typically has few values,
examples) had semantic and formal assignment. and yet is pervasive in one language and absent in
754 Gender, Grammatical

another, thus providing helpful control conditions. 22 morphological classes, and the latter are largely
Psycholinguists are also interested in apparently non- predictable from phonological information. It has
linguist effects of grammatical gender, as in judgments been demonstrated how this system can be analyzed
of picture similarity (see Boroditsky et al., 2003). within the Network Morphology framework, and an
There has been some interesting work on how chil- implementation has been provided to demonstrate
dren acquire gender systems by Mills (1986) and that these claims are indeed valid (Fraser and Corbett,
Müller (2000), among others. Such studies may also 1997). There is also a complex interaction of assign-
help us to see how such systems change over time, as ment to gender and to morphological class in Bininj
shown by the work of Polinsky and Jackson (1999) Gun-Wok (previously Mayali), a non-Pama-Nyungan
on Tsez (Dido); see also Comrie and Polinsky (1998). language of northern Australia, which has four gen-
For further work on modeling of change in assignment ders and five morphological classes for nouns. This
systems see Polinsky and Everbroeck (2003). An too has been successfully implemented; for details of
account of the history of gender in Indo-European is a formal model of this system see Evans et al. (2002).
provided by Matasović (2004). Modeling has tended to concentrate on synchronic
assignment systems. However, there is also interesting
Modeling work on modeling the development of assignment
systems over time, as in Polinsky and Everbroeck
Modeling is being used in various ways. First, it is
(2003), as noted earlier.
used to verify particular analyses. Some types of as-
signment system are so transparent that alternative
analyses are hard to imagine. However, with morpho- Conclusion
logical assignment systems, particularly when the
number of genders and the number of inflectional We have seen why gender continues to evoke such
classes are close in a given language, there are other interest: it is a core feature in some languages, while
possible analyses. Arguments in favor of the morpho- in many others it is absent. We have the analytic tools
logical approach include the following: the main for investigating gender systems, and new and in-
claim is that in languages like Russian, the internal triguingly complex systems are still coming to light.
evidence shows that we can assign gender from inflec- In terms of gender assignment, the rules vary from
tional class (but the inflectional class cannot be pre- being fully transparent, as in Bagvalal, to complex, as
dicted from gender); second, there is the typological in French and German. It is a research area in which
claim that one of the possible types of assignment various different types of expertise are leading to
system (or some combination) will always guarantee gradual progress.
that gender is predictable for the vast majority of
nouns, and so treating the Russian type system differ-
ently would create an anomaly. However, modeling
Acknowledgments
allows us to take a further step. Fraser and Corbett The support of the Economic and Social Research
(1995) employed the Network Morphology frame- Council (UK) under grant RES051270122 is grateful-
work to give an account of the Russian gender and ly acknowledged.
inflectional class relations. Since the analysis was
implemented in the lexical knowledge representation See also: Number; Phonology: Overview.
language DATR (Evans and Gazdar, 1996), the
analyses can be verified computationally. It can be
demonstrated that, starting from minimal lexical Bibliography
entries for a wide range of nouns, this account did
indeed predict the right genders. Aikhenvald A Y (2003). A grammar of Tariana, from
A more complex situation is found in languages Northwest Amazonia. Cambridge: Cambridge University
Press.
where gender is assigned according to morphological
Aronoff M (1992). ‘Noun classes in Arapesh.’ In Booij G &
class, but where the morphological class is itself pre-
van Marle J (eds.) Yearbook of morphology 1991.
dictable. A well-studied case is Bumbita Arapesh, a Dordrecht: Kluwer. 21–32.
language of the Toricelli family spoken on the north Aronoff M (1994). Linguistic inquiry monograph 22:
coast of New Guinea (Fortune, 1942; Aronoff, 1992; Morphology by itself: stems and inflectional classes.
1994: 89–114). In addition to semantic assignment Cambridge, MA: MIT Press.
rules based on sex, there are morphological assign- Austin P (1981). A grammar of Diyari, South Australia.
ment rules that determine 13 genders on the basis of Cambridge: Cambridge University Press.
 Gender, Grammatical 755

Bani E (1987). ‘Garka a ipika: masculine and feminine Kilarski M (2003). ‘Gender assignment in Danish, Swedish
grammatical gender in Kala Lagaw Ya.’ Australian Jour- and Norwegian: a comparison of the status of assignment
nal of Linguistics 7, 189–201. criteria.’ NORLYD 31(2): The Proceedings of the 19th
Boroditsky L, Schmidt L & Phillips W (2003). ‘Sex, syntax Scandinavian Conference of Linguistics. 261–274.
and semantics.’ In Gentner D & Goldin-Meadow S (eds.) Köpcke K M & Zubin D (1996). ‘Prinzipien für Genuszu-
Language in mind: advances in the study of language and weisung im Deutschen.’ In Lang E & Zifonoun G (eds.)
thought. Cambridge, MA: MIT Press. 61–80. Deutsch typologisch: Jahrbuch des Instituts für deutsche
Comrie B & Polinsky M (1998). ‘Gender in a historical Sprache 1995. Berlin: de Gruyter. 473–491.
perspective: radial categories meet language change.’ In Lindström E (2002). ‘Topics in the grammar of Kuot: a non-
Justus C F & Edgar C P (eds.) Journal of Indo-European Austronesian language of New Ireland, Papua New
Studies monograph 31: Language change and typological Guinea.’ Ph.D. diss., Stockholm University.
variation: in honor of Winfred P. Lehmann on the occa- Matasović R (2004). Gender in Indo-European. Heidel-
sion of his 83rd birthday: II: grammatical universals and berg: Winter.
typology. Washington, DC: Institute for the Study of Mel’nikov G P & Kurbanov A I (1964). ‘Logičeskie osno-
Man. 566–589. vanija imennoj klassifikacii v caxurskom jazyke.’ In
Corbett G G (1991). Gender. Cambridge: Cambridge Uni- Ivanov V V (ed.) Voprosy struktury jazyka. Moscow:
versity Press. Nauka, Moscow. 157–170.
Corbett G G (2005). ‘The number of genders,’ ‘Sex-based Mills A E (1986). The acquisition of gender: a study of
and non-sex-based gender systems,’ and ‘Gender assign- English and German. Berlin: Springer-Verlag.
ment systems’ [three chapters and maps]. In Haspelmath Müller N (2000). ‘Gender and number in acquisition.’ In
M, Dryer M, Gil D & Comrie B (eds.) World Unterbeck B, Rissanen M, Nevalainen T & Saari M (eds.)
atlas of language structures. Oxford: Oxford University Trends in linguistics: studies and monographs 124:
Press. Gender in grammar and cognition. Berlin: Mouton de
Corbett G G (forthcoming). Agreement. Cambridge: Cam- Gruyter. 351–399.
bridge University Press. Munro P (1998). ‘The Garifuna gender system.’ In Hill J H,
Cowell M W (1964). Arabic series 7: A reference grammar Mistry P J & Campbell L (eds.) Trends in linguistics:
of Syrian Arabic (based on the dialect of Damascus. studies and monographs 108: The life of language: papers
Washington, DC: Georgetown University Press. in linguistics in honor of William Bright. Berlin: de
Dixon R M W (1972). The Dyirbal language of North Gruyter. 443–461.
Queensland. Cambridge: Cambridge University Press. Parker E M & Hayward R J (1985). An Afar-English-
Evans N, Brown D & Corbett G G (2002). ‘The semantics French dictionary (with grammatical notes in English).
of gender in Mayali: partially parallel systems and formal London: School of Oriental and Asian Studies, University
implementation.’ Language 78, 111–155. of London.
Evans R & Gazdar G (1996). ‘DATR: a language for lexical Polinsky M & Jackson D (1999). ‘Noun classes: language
knowledge representation.’ Computational Linguistics change and learning.’ In Fox B A, Jurafsky D & Michaelis
22, 167–216. L (eds.) Cognition and function in language. Stanford:
Fortune R F (1942). Publications of the American Ethno- CSLI. 29–50.
logical Society 19: Arapesh. New York: J. J. Augustin. Polinsky M & Van Everbroeck E (2003). ‘Development of
[Reprinted 1977, New York: AMS Press.] gender classifications: modeling the historical change
Fraser N M & Corbett G G (1995). ‘Gender, animacy from Latin to French.’ Language 79, 356–390.
and declensional class assignment: a unified account for Reid N (1997). ‘Class and classifier in Ngan’gityemerri.’
Russian.’ In Geert B & van Marle J (eds.) Yearbook of In Harvey M & Reid N (eds.) Studies in Language Com-
morphology 1994. Dordrecht: Kluwer. 123–150. panion Series 37: Nominal classification in Aboriginal
Fraser N M & Corbett G G (1997). ‘Defaults in Arapesh.’ Australia. Amsterdam: John Benjamins. 165–228.
Lingua 103, 25–57. Seifart F (2004). ‘Nominal classification in Miraña, a
Hardison D M (1992). ‘Gender assignment to nonwords in Witotoan language of Columbia.’ Sprachtypologie und
French: implications for the role of the final syllable Universalienforschung 57, 228–246.
in lexical processing and organization of the mental lexi- Taylor D (1977). Languages of the West Indies. Baltimore:
con.’ In Indiana Linguistics Club twenty-fifth anniversary Johns Hopkins University Press. [cited from Munro,
volume. Bloomington, IN: Indiana University Linguistics 1998.]
Club. 13–25. Tucker G R, Lambert W E & Rigault A A (1977). The
Hockett C F (1958). A course in modern linguistics. New French speaker’s skill with grammatical gender: an exam-
York: Macmillan. ple of rule-governed behavior. The Hague: Mouton.
Ibragimov G X (1990). Caxurskij jazyk. Moscow: Nauka. van Berkum & Jos J A (1996). The psycholinguistics of
Kibrik A E (1999). Èlementy grammatiki caxurskogo jazyka grammatical gender: studies in language comprehension
v tipologičeskom osveščenii. Moscow: Nasledie Press. and production. Nijmegen: Nijmegen University Press.
Kibrik A E (ed.) (2001). Bagvalinskij jazyk: grammatika: Zaliznjak A A (1964). ‘K voprosu o grammatičeskix kate-
Teksty: Slovari. Moscow: Nasledie. [co-editors Kazenin gorijax roda i oduševlennosti v sovremennom russkom
K I, Ljutikova E A & Tatevosov S G.] jazyke.’ Voprosy jazykoznanija 4, 25–40.
756 Gender, Grammatical

Zubin D & Köpcke K (1986). ‘Gender and folk taxonomy: ings of a symposium on categorization and noun classifi-
the indexal relation between grammatical and lexical cation, Eugene, Oregon, October 1983. Amsterdam:
categorization.’ In Craig C (ed.) Typological Studies in John Benjamins. 139–180.
Language 7: Noun classes and categorization: proceed-

Gender, Sexuality and Language

M Bucholtz, University of California, Santa Barbara, indexing interactional stances such as affect or force/
Santa Barbara, CA, USA mitigation, not gender identities.
K Hall, University of Colorado, Boulder, Despite its flaws, the early anthropological re-
CO, USA search on women’s and men’s languages did call at-
ß 2006 Elsevier Ltd. All rights reserved. tention to the important relationship between gender
and sexuality. However, this connection was not de-
veloped theoretically until the 1990s, in spite of the
Gender initially became a focus of linguistic inter- small but steady stream of linguistic publications on
est in the first half of the 20th century when field sexuality. In the 1970s, a number of ethnographically
linguists discovered what they perceived as stark oriented researchers published a flurry of studies on
differences between European languages and the in- sexualized insults and banter, focusing primarily on
digenous languages they had begun to document in male speakers. Although the intention of such work
the Americas and elsewhere. Of particular interest was to bring underinvestigated communities and gen-
was the finding that a number of these languages res into linguistic scholarship, many of these studies
differentiated between women’s and men’s speech unwittingly worked to reinscribe stereotypes of the
on the basis of grammar, phonology, and lexicon. licentious and hypersexed Other.
These so-called ‘women’s languages’ and ‘men’s As linguistic anthropologists began to turn their
languages’ were characterized as vastly different and attention to at least some aspects of sexuality, gender
mutually exclusive, and were often held up as evi- was gaining a more central role in linguistic scholar-
dence for the rigidity of gender roles in traditional ship due to the influence of second-wave feminism,
societies in contrast to the enlightened gender liberal- especially in sociolinguistic research on Western lan-
ism of Western modernity. In addition to the prob- guages. One of the earliest and most important con-
lematic exoticism of native languages and cultures tributions to this new line of feminist scholarship
that informs this view, such a dichotomy between offered a different conceptualization of ‘women’s lan-
traditional and modern structures of gender is guage’ as primarily pragmatic rather than structural
empirically untenable. It has recently been shown, and suggested that women’s speech both produced
for instance, that it was only under the conditions and reflected real-world powerlessness. This view was
of modernization that gender differentiation through complicated by contemporary anthropological re-
the creation of ‘women’s language’ emerged in Japan. search, however, which revealed that women’s speech
In these early texts, sexuality was not theoret- in other cultures could be forceful and assertive, though
ically distinguished from gender; researchers assumed still devalued.
a direct mapping from one to the other and of both Recognizing the widespread deprecation of social
onto language. Thus a speaker’s departure from practices associated with women, by the 1980s many
normative speech patterns was interpreted as gender feminist social scientists, including linguistic anthro-
deviance as well as sexual deviance, with ‘effeminate’ pologists, were seeking to validate women as compe-
and ‘bisexual’ speakers occupying the margins as tent cultural members in their own right. Meanwhile,
linguistic exceptions to an otherwise unyielding the concept of ‘culture’ was adopted by some re-
gender dichotomy. What was missing from such a searchers of American society to account for what
perspective was the concept of ‘indexicality,’ the pro- were still seen as vast differences between women
cess whereby language ‘points to’ the social and dis- and men. Though inspired by anthropological work
cursive context of its own production. Seen in this on cross-cultural communication, this body of work
way, many instances of perceived cross-gender lan- overlooked widespread efforts to rethink the concept
guage use might more accurately be understood as of culture within anthropology in this period.