Grammatical Gender Research Paper

1. Terms

The defining characteristic of gender is agreement; a language has a gender system only if noun phrases headed by nouns of different types control different agreements. The evidence that nouns have gender in a given language thus lies outside the nouns themselves. There is no substantive difference between ‘genders’ and ‘noun classes’; the different terms are merely the products of different linguistic traditions. For instance, we find systems with three genders, to which nouns are assigned by similar rules, in both Kannada (a Dravidian language) and in Godoberi (a Nakh-Daghestanian language). By tradition the first is said to have three genders, and the second three noun classes. We shall use the term ‘gender’ in both instances.

2. The Number Of Genders

The approach to gender sketched in Sect. 1 is based on Zaliznjak’s (1964) notion of ‘agreement class.’ Two nouns are in the same agreement class only if they take the same agreements under all conditions, that is if we control for other grammatical categories such as case and number. Conversely, if two nouns differ in their agreements when factors such as case and number are held constant, then they belong to two different agreement classes and it will normally be the case that they belong to two different genders. While in many languages there is no dispute as to the number of genders, there are other languages where the question is far from straightforward. In the case of Romanian there has been a lengthy debate. The analytical problem of determining the number of genders and the tests for deciding the gender of a given noun depends on separating out the classes into which nouns are divided (the controller genders) from the number of different genders marked on agreement targets (the target genders). Frequently the two match up, but in several languages they do not. A full treatment of the subject is not possible here; for a detailed account and extensive references see Corbett (1991).

3. Gender Assignment

We now ask how nouns are distributed over the genders of a given language, in other words, how does a native speaker ‘know’ what gender a noun belong to? Models of the mechanisms by which nouns are allotted to genders are called ‘assignment systems.’ Assignment may involve two sorts of information about the noun: its meaning and its form. We shall consider semantic systems first.

3.1 Strict Semantic Assignment Systems

This type is found in Dravidian languages like Kannada, spoken in the state of Karnataka, southern India (Sridhar 1990, p. 198). Nouns denoting male humans are masculine, those denoting female humans are feminine. There are also deities, demons, and heavenly bodies in these genders. All remaining nouns, including those denoting infants and animals, are neuter. Thus appa ‘father,’ and candra ‘moon’ are masculine, amma ‘mother’ is feminine, and na:yi ‘dog’ is neuter. Certain other Dravidian languages like Kolami have only two genders: nouns denoting male humans are masculine and all others belong in the nonmasculine gender (Emeneau 1955, p. 73). Elsewhere we also find the reverse. Thus in Diyari, a language of South Australia, there is a gender for nouns with female referents (such as women, girls, doe kangaroos), and the other is for all remaining nouns (Austin 1981). In Alamblak, a Sepik Hill language of Papua New Guinea, the masculine gender includes nouns denoting males and those denoting things like crocodiles, pythons, and arrows, which are long and thin, while the feminine is for nouns denoting females or short, squat items like turtles, frogs, and chairs (Bruce 1984, pp. 96–8, Foley 1986, pp. 80–81). An interesting system with some similar features is found in Taiap (Kulick and Stroud 1992).

3.2 Predominantly Semantic Assignment Systems

Many languages have semantic assignment rules, which, however, lack the coverage of those seen in Kannada. An example of what we shall call predominantly semantic assignment is found in Tsakhur, which is a member of the Lezgic group of the Nakh-Daghestanian family. Its four genders are shown in Table 1.

Assignment to genders I and II is straightforward: I is for male humans (also gods, angels, and so on) while II is for female humans (and female mythical beings). Most remaining animates are assigned to gender III, though a few belong to gender IV, along with some mythical beings. Inanimates are found in genders III and IV (Melınikov and Kurbanov 1964, Ibragimov 1990, pp. 54–6, and references therein, Kibrik 1999, pp. 48–9).

Languages of this general type are widespread. For some of them, researchers have proposed that abstract semantic criteria partly miss the point, and that if we can gain a better grasp of the world view of the speakers, we can then understand the assignment system. The most discussed case is that of Dyirbal, a language of North Queensland (Dixon 1972), which has four genders, primarily for: male humans and non- human animates (I), female humans (II), non-flesh food (III) and the residue (IV). There are numerous apparent exceptions: the moon is in the first, masculine gender and the sun is the second, feminine gender. However, in Dyirbal mythology, as indeed in much of Australia, the moon is the husband of the sun; in Dyirbal the mythological significance of referents is a key factor in assignment. Dixon’s data and analysis continue to provoke debate, as shown by the recent paper by Polinsky (1998). There is similar debate as the role of world view in the gender assignment system of Ojibwa (and in other Algonquian language of southern Canada and northern USA).

In all the languages discussed so far, the meaning of the noun determines gender. In the strict assignment systems, the rules are immediately obvious; in the predominantly semantic systems, there may be a minority of exceptions, but these exceptions have been claimed to be largely apparent in some languages, once the cultural setting of the language is taken into account. If we ask which are the semantic criteria on which semantic systems can be based, we see recurring patterns. Quite often we find animate inanimate, human nonhuman, and male female. Sometimes there is a gender for diminutives, as in various Bantu languages. There are also less usual genders, such as that for nonflesh food (Dyirbal) and the gender for insects (found in the Rikvani dialect of the Nakh-Daghestanian language Andi). A criterion which is sufficient to define a gender in one language may be just one factor in the assignment to a gender in another. Thus the Bantu language Chichewa has a gender for diminutives, while in Dizi, diminutives together with nouns denoting females form a gender.

3.3 Formal Assignment Systems

In many languages, assignment by semantic rules would fail to assign many nouns to a gender. While in languages like Kannada the nouns not assigned by the semantic rules (the ‘remainder’ or ‘semantic residue’), all belong to a single gender, in many languages they are spread over more than one gender. In such languages we find additional rules for assigning nouns to genders according to their form. Note the asymmetry: languages may use semantic rules, or semantic and formal rules, but not just formal assignment rules. In no language are nouns assigned to genders, as defined earlier, by purely formal rules. An example would be a language in which there were two agreement classes, and the nouns in the first all ended in a consonant cluster, and those in the second did not, and there was no semantic regularity for the distribution of nouns. I claim that this hypothetical type does not exist. Formal assignment rules may appeal to two types of information: phonological and morphological.

3.3.1 Phonological. A clear example of assignment depending on phonological information is provided by Qafar (Afar), an East Cushitic language spoken in north-eastern Ethiopia and in Djibouti (data 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. It is the nouns which fall outside these rules, the residue, which are of interest. For them the phonological rules apply: nouns whose citation form ends in an accented vowel are feminine (for example, karma ‘autumn’) while all others are masculine (for example, gilal ‘winter,’ which does not end in a vowel, and tamu ‘taste,’ which does end in a vowel, but not an accented one). These rules operate with few exceptions. Moreover, nouns denoting males and females typically fit them too (for example baqla ‘husband’ and barra ‘woman, wife’). It might seem that we could dispense with semantic rules for Qafar, and treat it as being of a quite different type. However, while the phonological rule works for the vast majority of cases, we find the crucial example abba ‘father,’ which is masculine, even though it ends in an accented vowel, which ‘should’ be an indicator of feminine gender. Conversely, gabbixeera ‘slenderwaisted female’ is feminine, though the accent is nonfinal. Qafar is remarkably for the simplicity of its phonological assignment rules, and for the fact that when the semantic and phonological rules are both applicable, they almost always give the same result. However, in cases of conflict, the semantic rules take precedence (as is the normal situation in gender assignment systems).

Other examples can occur in various parts of the world. Even French, often claimed to have no system to its gender assignment, has been demonstrated to have a phonological system, though not as simple as that of Qafar. For example, of 938 nouns ending in ε, 99 percent are masculine (like le pain [pε] ‘the bread’). And of 1,453 nouns in /Ʒ/ , 94.2 percent are masculine (like le menage [mena: ] ‘the household’ (for details see Tucker et al. 1977).

3.3.2 Morphological. In Russian, which is typical of many other Indo-European languages, nouns denoting male humans are masculine and those denoting female humans are feminine. But unlike the situation in languages like Kannada, it is not the case that the semantic residue is assigned to the neuter gender; in Russian the residue is shared between the three genders, with the neuter gender not even receiving the majority. Searching for additional semantic criteria is not at all promising, as the following data suggests (Table 2).

If we look instead at the morphology of the nouns, then we can make progress. There are four main inflectional classes in Russian, each with several thousands of nouns (for justification of this not uncontroversial view see Corbett 1982, pp. 202–11). There are six cases and two numbers (though no paradigm has 12 distinct forms because of various syncretisms). We give just the singular forms in Table 3.

The native speaker needs to know how a noun inflects, in order to produce grammatical utterances. On the basis of that information, the assignment rules are straightforward. Nouns in class I are masculine, those in classes II and III are feminine, and those in IV are neuter. At first we might think that the semantic assignment rules are superfluous, since mal’cik ‘boy’ is in class I, while de uska ‘girl’ is in class II, and mat ‘mother’ is in class III. In other words, many of the sex-differentiable nouns would be assigned to the appropriate gender by the morphological assignment rules. But there are also instances where this is not so, for instance, djadja ‘uncle,’ which denotes a male but is in class II, whose nouns are typically feminine. Djadja ‘uncle’ is masculine. Nouns like this show, once again, that we do not find languages where formal assignment rules are sufficient. Further rules are required in Russian for indeclinable nouns (like taksi ‘taxi,’ which is indeclinable and neuter). Morphological assignment systems are found in various other Indo-European languages. Looking further afield, in Arabic too, gender is assignable in the main according to morphology (Cowell 1964, pp. 372–5).

The different types of assignment criteria may overlap in ways which make it difficult to establish their relative weight in a given language. Thus there may be small clusters of nouns which can be accounted for by semantic criteria (apart from those covered by the main semantic rules) even within systems where formal rules have a major role (as shown for German by Zubin and Kopcke 1986, Kopcke and Zubin (1996). However, whenever gender languages are analyzed in sufficient detail, the gender of the vast majority of nouns turns out to be predictable.

4. Prospects

Studies on grammatical gender continue to appear regularly, including valuable studies of previously unresearched languages. Here we highlight two main areas where we can expect substantial new developments: lexicology and modeling.

4.1 Lexicology

Gender is a key category for lexicology, since it is crucial for understanding the nature of lexical entries. In a gender language the gender of a noun must be available. On the other hand, as we have seen, it does not normally need to be specified, since the information can be derived, by assignment rules, from other information which must be stored. We still do not know the full range of information which may be relevant (for instance, the possible semantic features which may determine gender assignment).

There are further surprises: it is possible too for certain nouns to be of different genders according to the sex of the speaker; this is found in Garifuna, a member of the Arawak family spoken in Belize, Honduras, Nicaragua, and Guatemala (Taylor 1977, p. 60, Munro 1998). These two points raise interesting questions as to the shape of lexical entries, a general question which is of concern to lexicologists and increasingly to psycholinguists (see, for instance, Berkum (1995).

4.2 Modeling

When discussing some types of assignment system, the evidence is so clear-cut that alternative analyses are hard to imagine. However, with morphological assignment systems, particularly when the number of genders and the number of inflectional classes is similar in a given language, there are other possible analyses, in particular some would specify gender in the lexical entry and derive inflectional information from it. Arguments in favour of the approach taken in Sect. 3.3.2 include the following; first, and primarily, it is claimed that in languages like Russian the internal evidence shows that the gender from inflectional class approach is preferable (it proves impossible to predict inflectional class from gender but it is possible to predict gender from inflectional class); and second, there is the typological claim that one or other of the possible types of assignment system (or some combination) will always be in play in a gender language such that gender is predictable for the vast majority of nouns and so if the Russian type system were treated differently this would create an anomaly.

However, there is a further step which we can take, of a rather different nature. Since the Russian type languages are crucial, it worth investing additional time in clarifying them. Hence in Fraser and Corbett (1995) we give an account of the Russian gender and inflectional class relations in the Network Morphology framework. The analysis is implemented in the lexical knowledge representation language DATR (Evans and Gazdar (1996, Gazdar forthcoming), which means that the analyses can be verified by computer. Given minimal lexical entries for nouns our account does indeed predict the right genders for a wide range of nouns. An account like this which is both explicit and testable is not available for the alternative approach.

A more complex type is that in which gender is again assigned according to morphological class, but where the morphological class is itself predictable. One such case is Arapesh; a language of the Torricelli family, spoken on the north coast of Papua New Guinea. Genders and morphological classes form a substantial part of Fortune’s (1942) grammar, which led to work based on it by Aronoff (1992, 1994, pp. 89–114). Once again, there are semantic assignment rules based on sex. But then there are morphological assignments rules, which determine 13 genders on the basis of 22 morphological classes (which are in turn predictable largely from phonological information). Again we have demonstrated how this system can be analyzed within the Network Morphology framework and have provided an implementation to demonstrate that our claims are indeed valid (Fraser and Corbett (1997). There is a complex interaction of assignment to gender and to morphological class in Mayali, a non-PamaNyungan language of north Australia, which has four genders and five morphological classes for nouns. For the essential data see Evans (1997) and for details of a formal model of this system see Evans et al. (2001). Finally, there is interesting work on the development of assignment systems over time, including some modelling: see Comrie and Polinsky (1998), Polinsky and Everbroeck (1998), Polinsky and Jackson (1999).

5. Conclusion

Gender remains a puzzling category, apparently deeply embedded into the structure of some languages, and missing in other. This is found even in related languages: thus several Indo-Aryan languages have two or three genders, with substantial agreement, while another, Bengali, has no gender even in the personal pronouns. Where we find gender, the assignment rules can be simple, as in Kannada and Qafar, or complex as in French and German. It is an area in which gradual progress is being made and one in which there is still much to learn.

Bibliography:

1. Aronoff Mark 1992 Noun classes in Arapesh. In: Booij G, van Marle J (eds.) Yearbook of Morphology 1991. Kluwer, Dordrecht, The Netherlands, pp. 21–32
2. Aronoff M 1994 Morphology By Itself: Stems and Inflectional Classes (Linguistic Inquiry Monograph 22). MIT Press, Cambridge, MA
3. Austin P 1981 A Grammar of Diyari, South Australia. Cambridge University Press, Cambridge, UK
4. Berkum J J A van 1996 The psycholinguistics of grammatical gender: Studies in language comprehension and production. Ph.D. dissertation. Nijmegen University Press, Nijmegen, The Netherlands
5. Bruce L 1984 The Alamblak Language of Papua New Guinea (East Sepik). Department of Linguistics, Research School of Pacific Studies, Australian National University, Canberra, ACT
6. Comrie B, Polinsky M 1998 Gender in a historical perspective: Radial categories meet language change. In: Justus C F, Polome E C (eds.) Language Change and Typological Variation: In Honor of Winfred P. Lehmann on the Occasion of His 83rd Birthday: II Grammatical Universals and Typology. Institute for the Study of Man, Washington, DC, pp. 566–89
7. Corbett G G 1982 Gender in Russian: An account of gender specification and its relationship to declension. Russian Linguistics 6(2): 197–232
8. Corbett G G 1991 Gender. Cambridge University Press, Cambridge, UK
9. Cowell M W 1964 A Reference Grammar of Syrian Arabic (based on the dialect of Damascus) (Arabic series 7). Georgetown University Press, Washington, DC
10. Dixon R M W 1972 The Dyirbal Language of North Queensland. Cambridge University Press, Cambridge, UK
11. Emeneau M B 1955 Kolami: A Dravidian Language. University of California Press, Berkeley and Los Angeles
12. Evans N 1997 Head classes and agreement classes in the Mayali dialect chain. In: Harvey M, Reid N (eds.) Nominal Classification in Aboriginal Australia. John Benjamins, Amsterdam, pp. 105–46
13. Evans N, Brown D, Corbett G G 2001 The semantics of gender in Mayali: Partially parallel systems and formal implementation. Language 77
14. Evans R, Gazdar G 1996 DATR: A language for lexical knowledge representation. Computational Linguistics 22: 167–216
15. Foley W A 1986 The Papuan Languages of New Guinea. Cambridge University Press, Cambridge, UK
16. Fortune R F 1942 Arapesh. J J Augustin, New York
17. Fraser N M, Corbett G G 1995 Gender, animacy and declensional class assignment: A unified account for Russian. In: Booij G, van Marle J (eds.) Yearbook of Morphology 1994. Kluwer, Dordrecht, The Netherlands, pp. 123–50
18. Fraser N M, Corbett G G 1997 Defaults in Arapesh. Lingua 103: 25–57
19. Gazdar G forthcoming Ceteris paribus. In: Kamp J A W, Rohrer C (eds.) Aspects of Computational Linguistics. Springer, Berlin
20. Ibragimov G X (ed.) 1999 Caxurskij jazyk. Nauka, Moscow
21. Kibrik A E (ed.) 1999 Elementy grammatiki caxurskogo jazyka tipologiceskom os escenii. Nasledie Press, Moscow
22. Kopcke K M, Zubin D 1996 Prinzipien fur Genuszuweisung im Deutschen. In: Lang E, Zifonoun G (eds.) Deutsch typologisch: Jahrbuch des Instituts fur deutsche Sprache 1995. de Gruyter, Berlin, pp. 473–91
23. Kulick D, Stroud C 1992 The structure of the Taiap (Gapun) language. In: Dutton T, Ross M, Darrell T (eds.) The Language Game: Papers in Memory of Donald C. Laycock. Department of Linguistics, Research School of Pacific Studies, Australian National University, Canberra, ACT, pp. 203–26
24. Mel’nikov G P, Kurbanov A I 1964 Logiceskie osnovanija imennoj klassifikacii v caxurskom jazyke. In: Ivanov V V (ed.) Voprosy struktury jazyka. Nauka, Moscow, pp. 157–70
25. Munro P 1998 The Garifuna gender system. In: Hill J M, Mistry P J, Campbell L (eds.) The Life of Language: Papers in Linguistics in Honor of William Bright. de Gruyter, Berlin, pp. 443–61
26. Parker E M, Hayward R J 1985 An Afar-English-French Dictionary (with Grammatical Notes in English). SOAS, University of London, London
27. Polinsky M 1998 Women are not dangerous things: Grammatical gender and categorization. Paper presented at the Workshop ‘Processing of Grammatical Gender,’ Max Planck Institute of Cognitive Neuroscience and Centre of Cognitive Sciences, University of Leipzig, Leipzig, 24–25 July 1998
28. Polinsky M, Everbroeck E van 2000 Noun classes: Language change and learning. Development of gender classifications: Modeling the historical change from Latin to French. Unpublished paper, UCSD
29. Polinsky M, Jackson D 1999 Noun classes: Language change and learning. In: Fox B A, Jurafsky D, Michaelis L A (eds.) Cognition and Function in Language. CSLI, Stanford, pp. 29–50
30. Sridhar S N 1990 Kannada (Descriptive Grammars Series). Routledge, London
31. Taylor D 1977 Languages of the West Indies. Johns Hopkins University Press, Baltimore, MD
32. Tucker G R, Lambert W E, Rigault A A 1977 The French Speaker’s Skill with Grammatical Gender: An Example of Rule-governed Behavior. Mouton, The Hague
33. Zaliznjak A A 1964 K voprosu o grammaticeskix kategorijax roda i odusevlennosti v sovremennom russkom jazyke. Voprosy jazykoznanija 4: 25–40
34. Zubin D, Kopcke K M 1986 Gender and folk taxonomy: The indexal relation between grammatical and lexical categorization. In: Craig C (ed.) Noun Classes and Categorization: Proceedings of a Symposium on Categorization and Noun Classification, Eugene, Oregon, October 1983. John Benjamins, Amsterdam, pp. 139–80