Version of 2008-11-26

Wersja polskaWersja dwujęzycznaBilanguage version

Grzegorz Jagodziński



  1. Introduction
  2. Nouns, substantives, adjectives
  3. What is gender
  4. What is not gender
  5. Similarities and dissimilarities between various types of nominal classes
  6. Nominal classes in various languages


Gender is a grammatical category. As it is totally absent in English, a lot of English-speaking linguists and enthusiasts of linguistics (who are evidently too lazy to study foreign and classical languages) have serious problems with understanding it correctly. As the result, they create bizarre theories, not compatible with the long tradition and sometimes even with common sense. The aim of this article is to show some of more common errors in understanding gender.

One important notice is needed before discussing the main subject. Namely, English-speaking authors, especially those from the U.S.A., are famous for making medleys of well-known notions and meanings. It is hard to explain why it is so. Perhaps some of them are simply undereducated, and write their papers not having understood the common use of notions they apply. Perhaps others mix meanings in purpose because they want to become famous by publishing a new hypothesis together with a new notional apparatus, made of old notions with new meanings. Perhaps they want to break the Greek-Latin tradition, and to replace it by a new one, their own, that is the American one. Whatever is the actual reason of the state of the art, the result is a growing mess in science. It also concerns linguistics.

Nouns, substantives, adjectives

Here is an example of the mess in lingustic notional apparatus caused by American authors. As it will appear in following parts of the article, the example is connected directly to the problem of gender. In the traditional grammar terminology which is continued in many European countries, nouns (Lat. nomina) are understood as declinable parts of speech, that is as substantives, adjectives, some numerals and some pronouns altogether. From the beginning of the 20th century, some American authors started to understand nouns as substantives only. It was a pity that some authors tried to break the nomenclatoric convention which had been living for a very long time. It is not known for what they needed it. Anyway, the new “style” of understanding nouns spread like the plague, and now some titled scholars in linguistic do not even realize that the word “noun” does not mean what they think it does. Evidently they have not read even one of older grammars written by their colleagues who lived only some tens of years ago. If they knew older bibliography, they would realize that, for instance, both words in “good man” should be termed nouns, in accordance with the Latin tradition, still present in almost all countries except the English speaking ones (which have followed the new bad American tradition). The first word of the given example (“good”) is called an adjective while the second one (“man”) is termed a substantive.

Unfortunately in many newer works only “man” is called a noun while “good” is not. This causes no common term for both substantives (incorrectly termed “nouns”) and adjectives. Indeed, these two classes of words (parts of speech) have little in common in English. But English is not a typical language in this point, and especially it differs from other Indo-European languages in which adjectives and substantives have much in common. The same is in many Uralic, Altaic or Afro-Asiatic languages. Even if joining substantives and adjectives in one superclass (of nouns) is not well-founded for each language of the world, also the new English-based style of emphasizing only differences between both classes is not well-founded. And it is especially groundless for most Indo-European (and Semitic) languages in which actual, especially morphological differences between substantives and adjectives may be really inconsiderable.

What is gender

The confusions with the notion of gender and misunderstandings concerning to it may be based on a similar reason, that is on breaking the good old tradition. Thus, let’s start from indicating what are the borders of the notion of gender.

Gender is selective for substantives (and sometimes for other parts of speech). Thus, in a language with gender, a given substantive is able to impose the proper form upon adjectives (numerals, pronouns) which determinate it (or upon pronouns which replace it – which however is less important). We say that adjectives and pronouns (in such languages) are declinable for gender. It means that it is necessary to apply various forms of adjectives and pronouns, depending on gender of the substantive which is determinated by them.

Polish is a good example of a language with gender. In this language, for example, stół ‘table’ is masculine, rzecz ‘thing’ is feminine, and biurko ‘desk’ is neuter. This is why one must say duży stół ‘big table’, duża rzecz ‘big thing’, duże biurko ‘big desk’, with various forms of the adjective, imposed upon by the gender of the substantive.

Moreover, the proper form of a pronoun is also required. Compare the following Polish and English examples:

It is hard to determine what makes a given substantive to be of such or another gender. In Polish in many instances there exist some formal features which can help in guessing the gender. Namely, most substantives with -a are feminine, most those with -o, -e, -ę are neuter, and finally most substantives which end in a consonant are masculine. Unfortunately, there are many exceptions: mężczyzna ‘man’, satelita ‘satellite’, książę ‘prince’, tato ‘dad’ are all masculine, noc ‘night’ is feminine etc. It may be specially interesting that gość ‘guest’ is masculine but kość ‘bone’ is feminine.

One can find almost none formal gender indicators in other languages, e.g. in German. In Latin there also exist some rules: substantives which end in -us, -er are masculine, in -a are feminine, in -um are neuter. But these rules are not strict (ex. humus ‘earth, soil’ is feminine and virus ‘venom’ is neuter), and many substantives have different endings than the listed ones, so you cannot guess their gender at all.

Thus, in general, no kind of “gender markers” (or more correctly: morphemes) on substantives are necessary, and substantives of various genders may look similar.

What is not gender

There is general agreement that genders are not the only possible classes of substantives – thus it is not true that we always have a gender system if substantives influence the form of anything else in the sentence. In various languages, substantives may be classified according to various criteria. In morphologic typology, not less than four groups of languages are distinguished (Majewicz 1989, see Bibliography):

  1. languages which have the category of gender (most Indo-European and Afro-Asiatic languages),
  2. classificatory languages (languages with noun classes: Bantu, some American Indian, some Caucasian languages),
  3. languages which have classifiers (Japanese, Chinese),
  4. languages which do not belong to any of these three groups (i.e. without gender, noun classes and classifiers, e.g. English or Hungarian).

The first three groups have various nominal classes (in a wider sense; they should not be confused with Bantu-like noun classes). Despite of the above classification, some English-speaking authors term most or even all of them genders. This is incorrect not only because of the tradition but also because it destroys boundaries between various classes of phenomena. In other words, the result of terming, for instance, Bantu classes genders is that we stop seeing differences between them.

There are also nominal classes which are not present in the above classification. For example, Hungarian is known for having no gender, no noun classes and no classifiers at all. Even the Hungarian pronoun ő means ‘he’ / ‘she’ / ‘it’ without difference. But there are distinctions between ki ‘who’ and mi ‘what’ and between hányan ‘how many people’ and hány ‘how many things’ there. Similar phenomena may be present in each single language. Nobody is likely to term them genders.

So, what is gender linked to and what is not? First of all, as we have already noticed, it is not determined by any markers. Besides, one should not think that Indo-European genders have too much to do with sex: “masculine” / “feminine” are just terms. In Polish, where genders exist, we cannot see such a link. Inanimate objects may be of masculine, feminine or neuter gender without any connection with their meanings. Moreover, sometimes gender is different than suggested by the meaning: dziewczę ‘little girl’ is neuter even if it denotes a female person.

Thus, gender is not strictly dependent of meaning. Instead, it is highly conventional and unforeseeable. Of course, sometimes the meaning of a given word is enough to guess its gender (that’s why mężczyzna ‘man’ is masculine and kobieta ‘woman’ is feminine in Polish) but sometimes it is not enough.

The same can be observed in other languages with gender. German Mädchen ‘little girl’ is neuter as well, and even Weib ‘woman’ is neuter there! Old English wīfman ‘woman’ was masculine while wīf ‘wife’ – neuter. There is also no connection between genders of a given word in various languages. For instance, German Schrank ‘wardrobe’ is masculine while Polish szafa ‘t.s.’ is feminine.

Similarities and dissimilarities between various types of nominal classes

Here is a comparison of the three types of nominal classes that have been listed above:

  genders noun classes classifiers
Number of classes several teens hundreds
Class affixes on substantives not necessary obligatory absent (classifiers are words)
Class affixes on adjectives obligatory the same as on adjectives absent
Agreement of adjectives with the substantive present present absent
Nominal class in singular and plural as a rule, the same different no category of number
Types of declension several no declension types no declension types

As one can easily see from the above chart, genders, noun classes and classifiers are three separate phenomena, and the number of differences between them is bigger than the number of similarities. All of them are some classes of substantives. Declension types (like those present in Latin) are also such classes. We do not call them genders, and by analogy we should not treat all classes of substantives as a kind of genders. Unfortunately, some English-speaking authors do prefer such a groundless idea. The most serious mistake, occurring in works of those authors, inconsistant with tradition, is terming noun classes (present in Bantu languages) genders.

If we agree that a gender system causes an effect upon the choice of the form of something else in the sentence, Chinese and Japanese classifiers would not be part of a gender system since they do not influence anything else. But, on the other hand, we could notice that a given substantive needs the specified classifier, and then the system would be a sort of a gender system. As we can see, influencing something is not enough to be the correct criterion to aknowledge a category gender.

In modern Mandarin the classifier can sometimes change into a suffix. This concerns the “universal” classifier -ge which is used when it is hard to choose between other classifiers. It has no tone and, as a consequence, we should treat combinations of a numeral and the classifier as one word. For example, sì zhī qiānbǐ ‘four pencils’ does not affect anything else. The classifier zhī is a separate word, a part of neither the numeral nor the substantive qiānbǐ. But when we count people, we should not use any classifiers but a special form of the numeral (with the suffix -ge) instead: sìge rén ‘four persons’. So, substantives like rén ‘man / men’ do influence the form of the numeral (and ‘this’ / ‘that’ pronouns) and, as a consequence, we should talk about genders in Chinese. But we do not do so, and, as a consequence, we would need better arguments for terming Bantu noun classes genders that the fact that they affect something in the sentence.

In Vietnamese classifiers can be used even without numerals – and we can treat them as a part of the substantive itself. They affect nothing in the sentence. Besides, the Vietnamese system is very similar to Chinese and Japanese. We can say that each Vietnamese substantive belongs to one of hundreds of classes, and the only result of this is the choice of a classifier which is needed in most cases. But the classifier itself affects nothing more. Thus Vietnamese classifiers are exactly like Latin suffixes which affects nothing more (but the declension type) and which are not told to be parts of any gender system.

Because of the differences it is just to treat gender and classifiers as two different phenomena. And what about gender and noun classes?

One of American linguists said once: “gender and noun classes are the same thing. Europeans who thought that gender had to have something to do with sex called the Bantu genders «noun classes».” This quotation is nothing more than a proof of lack of knowledge of this linguist. First, gender were never believed to be anyhow related to sex. They always were two different terms not related to each other, cf. Latin genus and sexus, and only quite not long ago sex and gender mixed together in the American dialect. Every linguist should know about it. Second, there are much more serious differences between genders and noun classes than those related to sex.

Namely, in languages with classes (like Bantu):

This all contrasts strongly with the state of the art in languages with gender where:

We can say that the category of class in Bantu replaces both the category of gender, the category of number and the category of case in languages with gender. So, actually, there are very few similarities between Bantu and Polish / German / Latin systems, and both types should be strictly separated from one another, and they should not be treated as just varieties of gender systems. This conclusion is compatible with tradition, and opposes trials of some authors who – for reasons known only to them – do not want to see the listed differences, and prefer to join classes and genders within one category (called “gender” by instigators for making even more mess).

It is obvious that the categories of gender and of noun classes have common features – but we could say the same about gender and classifiers. Nevertheless terming them all genders sounds like a terminological nonsense: gender is a well defined notion and there is no need to spread it onto similar but different categories.

Beside the already mentioned differences there also exist other reasons for treating gender, noun classes and classifiers separately. These reasons are of typological nature, and will be discussed below.

Nominal classes in various languages

Refusing of existence of grammar categories in isolating (highly analytic) languages means nothing more but very shallow understanding of grammar. In fact, various categories can occur in languages of various types. In inflexional and agglutinating languages grammar categories manifest themselves in various word forms while in isolating languages they occur only in syntactic constructions with miscellaneous words (with auxiliary ones or even with words with full meaning). The difference is, however, not so big as someone could think. Namely in all languages a given substantive causes a specific morpheme to occur. While this morpheme is a part of a word in inflexional languages, it remains a separate word in isolating languages.

Various types of nominal classes (not necessarily genders) are widely spread. Classifiers (mainly numeral classifiers) are present in East and South-East Asian languages of various families, like Chinese, Ainu, Japanese, Korean, Burmese, Thai, Hmong, Vietnamese, Munda, Bengali – but also in Mayan, Tsimshianic, Yagua (South America, the Peba-Yaguan family) and Georgian. Noun classes are present not only in Bantu; among others, they also occur in North Caucasian languages. Some Australian languages have four nominal classes. Algonquian languages use the binary category of animacy. An interesting instance of nominal classes can be found in Kiowa (a North American language from the Kiowa-Tanoan family) which has four classes of substantives based on inherent number (which do not show up besides): the class I comprises inherently singular / dual substantives, the class II – dual / plural ones, the class III – dual ones; class IV substantives do not distinguish number (they are noncount substantives). Any substantive used in non-inherent value of number takes the suffix -gau.

Gender is not a phenomenon commonly spread among the languages in the world. It occurs in all Semitic and Indo-European languages with only few exceptions (one of them is English). It is correct to think that English does not possess the category of gender (which is rather unique among Indo-European languages) as in this language there are no gender-dependent adjective forms. The only distinctions show up among pronouns which replace substantives – but it is not enough to speak of gender. Moreover, the choices between pronouns: personal he – she – it, interrogative who – what and relative who – which depend on meaning and as such, they cannot be taken as a proof. Similar distinctions are present in all languages of the world. For example Hungarian, which is generally accepted as not having gender, has also different forms for ‘who’ – ki and ‘what’ – mi, also used in the relative function (together with aki ‘who’ and ami ‘which’). Finally, the he – she – it system is not fully compatible with the who – what and who – which distinctions.

English had the category of gender once, and we can still observe few remnants of the previous state: elephant, bear, ox, horse, dog, love, fear, anger, war, time, day, death, ocean, winter, autumn, summer, wind, hurricane, sun as well as names of rivers and mountains are sometimes treated like masculine while cat, mouse, ship, yacht, man-of-war, car, aeroplane, locomotive, nature, earth, justice, fame, fortune, night, morning, liberty, victory, moon as well as names of countries (esp. England, the United States) and cities are treated like feminine – i.e. they are replaced by he or she respectively, instead of it. As it was stated, it is not enough to say about gender. Moreover, all the mentioned substantives may also be treated as neutral (may be replaced by it), and are often treated as such in normal speech. The hesitation in using various pronouns is illustrated by the following anecdote:

Breakfast table, a wasp hovering about the jam.
— Lady: ‘Look at him’.
— Gentleman (foreigner): ‘Why do you call it him?’
— Lady: ‘Because we look at it as something horrible’.

(after G. Wendt, Syntax des heutigen Englisch, Heidelberg 1911, part 1, p. 100.)

A little introduction to the problem. Nowadays the notion of “gender” is often understood inconsistantly with its traditional meaning – namely it is changed (mainly expanded) too much, and as a result, it begins to comprise phenomena (various nominal classes) which have nothing to do with gender when understood traditionally. This causes us to stop seeing things which are recognizable without all those changes. The first change mixes gender and sex (probably because of so called political correctness which makes people speak “gender” instead of “sex” – in fact those two notions are completely incompatible). Even if we have, as a rule, masculine and feminine among genders in Indo-European or Semitic languages, they have actually little to do with sex (e.g. many abstract things in Latin are feminine, not neuter, Old English wifman “woman” was masculine while wif “wife” was neuter; more similar examples can be found among virtually all IE languages).

The second distortion is to speak of gender every time when a noun affect virtually anything in a sentence, like relative pronouns. This is why some speak of animate and inanimate genders in English. The first of them as if demands using “who” while the other – “which”. If we understand gender this way, the phenomenon that occurs in Latin or German, and which is different from the who/which matter, would simply be left without a term. Moreover, the difference “who” – “what” is probably preserved in all languages, so linking it to gender is simply unpractical. There are yet other “popular” expansions of the range of the notion of gender but there are little related to Kiowa.

In my understanding (which is, I believe, close to the original definition), gender is always only a selective category. Its most important features are: there are no gender markers on nouns (contrary to noun classes known from Bantu), gender changes are rare and never applied for building case forms (once again, contrary to Bantu noun classes), and the whole classification is based only on various forms of adjectives (more precisely: words which are subordinate to nouns). (We can recognize gender from the form of adjective, not noun itself.) The category of (such defined) gender is present in Indo-European languages (English has no gender and it may even be the only exception among the whole family) and in Afro-Asiatic (anyway, in Semitic). The following Latin examples may serve as illustration: pulcher means “beautiful”, and indeed puer pulcher means “beautiful boy”. But in order to say “beautiful girl” we need another form of the adjective: puella pulchra. It is so because puella “girl” is of different gender than puer “boy”. We must also say puellum templum for “beautiful temple” as templum is of yet another gender. We see the same when taking the pronoun “this” into consideration: hic puer – haec puella – hoc templum.

Gender is a grammatical category (generally not related to meaning), and cannot be taken from the form of the noun – no gender markers exist. Even if -um nouns in Latin are neuter, -a nouns are, as a rule, feminine, while -us nouns – masculine, these are only loose indications, not gender morphemes (thus animal is neuter even without any marker, virus is also neuter even if with the “masculine marker”, humus is feminine, and poeta masculine). Once again, a similar situation is in all other IE languages (except English) and in many (all?) Afro-Asiatic ones. My general question is whether there exist languages with gender outside these two families. The available sources say little or none on this as their authors avail themselves different, too wide definitions of gender, not compatible with the one used by me (and in traditional grammars).

And finally, the concrete question on Kiowa. This language is famous of its 4 nominal classes and their unique and interesting features. Some time ago a certain linguist told me that Kiowa has 4 “genders” but his understanding of this notion is completely different than the standard one. But let’s leave all this terminological mess behind: my question is but very strictly formulated. Namely, the question is whether the state of the art in Kiowa resembles the one in Latin (as described above) or not. Is it true that the noun classification in Kiowa does not show up beside the meaning (sg – du – pl) of the basic and modified forms of a noun?