In languages with grammatical gender that has (almost) no morphological relation between the words and the genders(e.g. French), how do they determine the gender of a new word that has been introduced/created?

I think that as people don't speak based on the grammar but the grammar is based on how people speak, grammatical genders are determined based on in what gender people use the word.

But then, How do people know in what gender they are to treat the introduced/created word?

This depends a lot on the language, but if you know any example of any language, I'd be glad to know that.

  • 2
    I think that decide is not the right word. Jan 8, 2013 at 1:00
  • 3
    There has been a project on Loanword Typology: "we studied lexical borrowing patterns in 41 languages from around the world. ... For each language, we assembled lexical data for a fixed list of 1460 meanings." They produced a database. Their data can be useful for some testing of the hypothesized rules of adapting a word in languages with genders. Not sure whether they paid much attention to genders in their work. See also the book produced by the project, and email.eva.mpg.de/~haspelmt/lwt-guidelines.html to know which data was collected. Jan 9, 2013 at 18:35
  • If you find some conclusions about genders in their work, it would be nice if post your findings as an answer here, because it would be interesting to know. Jan 9, 2013 at 18:37
  • Have also a look at the search results for "Loanword Typology" gender. Jan 9, 2013 at 18:42
  • Wikipedia says that in Quebec French, words that end in consonant sounds in isolation are generally assumed to be feminine. In French, masculine nouns get their final consonant elided in most contexts. Jan 10, 2013 at 0:56

7 Answers 7


If the source language already has genders, then they will often be taken along to the host language, provided that people in the host language know a little bit about the source language. But there are countless other subtle, complex factors.

Dutch has a neuter article (het) and the fused/common gender masculine-femine (de). There are no definite rules for borrowed words, and we never learned any such rules. I would know if there were any, but no: genders are assigned subconsciously. For existing Dutch words, there are some rules, but I think perhaps half the words have genders that feel arbitrary (though fixed) to us. For example, a person will nearly always be masculine or feminine, and words on -heid are always feminine; but for most other words you just have to know. The difference between masculine and feminine only comes up with possessive adjectives (the fork and his?/her? prongs) and personal pronouns (not that fork: I saw her?/him? lying over there). Colloquially, we usually use masculine pronouns for de words; in writing, I often have to look up the gender of a word, unfortunately. In the south and in Belgium, I think they don't use masculine by default.

Because we have a word het huis in Dutch ("the house"), English safehouse will become het safehouse, as it is very similar. Usually the similarity is less direct; then a word with a similar meaning may be picked as the example. For instance, suppose that Japanese *inko means house (I'm just making this up), then I am more likely to say het inko.

However, a second factor is how the word sounds compared to words that sound remotely similar. Words on -o are probably more often masculine in Dutch, because of the many Italian and Spanish words on -o, like impresario, virtuoso, amigo, gringo, etc., so I am more likely to say de inko because of the sound.

Thirdly, I think many languages have a kind of "default" gender for (foreign) words. In German, I was taught that newly borrowed words are normally feminine. So it was easy to learn die CD and such (although, if this rule is valid in German, there must be many categories of exceptions). In Dutch, masculine and feminine have all but merged, that is, they have the same article, de. For that reason, de is much more frequent than het (neuter), so that de is more likely to be chosen for new words.

Fourthly, there is the possibility that the gender is taken from a third language, if that language has already adopted the word and assigned a gender to it. Dutch might follow German in some cases; I don't know how often this really happens.

To all these factors may be apportioned different weights in different circumstances, and they are somehow weighed by the subconscious of the first people who use the word, until at some point it has acquired a commonly accepted gender. It is often unpredictable in my experience.

Note also that it is perfectly possible for a borrowed word to remain of uncertain gender indefinitely. We even have some ancient native (inherited) words that we still bicker over (in a good-natured manner, of course). Now a few practical examples.

Dutch het net => het Internet. Since net means the same thing in Dutch as in English, this one was easy.

Het koekje => de/het cookie. I'm not sure about this one; I would probably say de cookie. Perhaps it is too new, but people do talk about this on a regular basis, have done so for at least a decade, and yet it is still not entirely decided. One would expect such a strong similarity to the Dutch word to lead to het cookie, and yet I would say de cookie. I feel that perhaps the default gender de is just too strong in Dutch. Or perhaps the fact that it is a small object: perhaps those have an extra strong tendency to be de words? Hard to say.

De router, de server: we have no similar words, but words on -er usually indicate an agent, a person even, and people are normally not neuter, so definitely de router, de server.

De absence: this is the French word, pronounced as in French. It was adopted into Dutch long ago, so I did not witness its introduction. My guess is that the ending -e had a strong influence, and simply the fact that it is obviously feminine in French.

Het communisme: this probably came from French. For some reason, (nearly?) all words on -isme are neuter in Dutch. I have no idea why, because I think they are masculine in all other languages, including French, which doesn't have a neuter gender at any rate. They come from Latin -ismus, which is masculine, from Greek -ismos, also masculine. Very strange.

De coating (as in paint). There is no Dutch equivalent. The ending -ing is nearly always feminine in Dutch for non-persons, so de (cognate to German -ung).

De axolotl: nothing at all sounds like or resembles an axolotl, but I think animals are by default de, because they are agents (even though we do have plenty of neuter animals).

De tsunami: probably de because -i sounds a bit like Italian or Latin masculine? There is also de golf "the wave", de vloed "the flood". I'm not sure.

Het DNA: hard to say. Perhaps because Acid was translated subconsciously as het zuur by the first Dutch scientists? Because substances are more likely to be het?

  • 1
    I had always been taught that German loan words default to neutral gender ("das") but clearly it's more complex than that. dartmouth.edu/~german/Grammatik/Wortbildung/Fremdwoerter.html
    – tripleee
    Jan 13, 2013 at 16:27
  • 5
    @tripleee: Perhaps the rule is, in the typical German fashion, "borrowed words are neuter, except... <20 pages of exceptions and sub-rules> ".
    – Cerberus
    Jan 13, 2013 at 17:16

As a native speaker, let me try to describe it. To be most specific, I will refer Ukrainian and Russian.

Short rules:

  1. Keyword: "a city of New York"; city is masculine, hence most often, city names are masculine as well;
  2. Pronunciation: words ending with consonant - masculine; ending with -e, -o - usually neutral; -a - feminine;

TL;DR, just if you are curious to details

First, let me remind that grammatical gender (GG) has no relation to lexical meaning of the word; you have just to remember GG of each noun.

Also, although there are three formal GG, there are nouns that are always singular, plural or dual.

Most loanwords belong to the following groups:

  • No declension: coffee, cocoa, Bourdeau (wine), Euro (currency), etc;
  • Toponyms: America, Tokyo, Mississippi, Himalaya;
  • Abbreviations (OUN, USA);
  • Noun groups
  • Naturally plural or paired items (types of shoes);

General rule

  • Those ending on -o, -e (many French and Italian words) usually become neutral;
  • Ending with -a become feminine;
  • Ending with a plosive consonant become masculine;
  • Ending with "soft sign" (pronounced with a palatalized final consonant) so feminine;

Nouns that do not decline

They usually used to be proper nouns that later started to mean common terms (whisky, Champagne, Bourdeau, brandy, Euro). Note they have retained capitalization in English, but in Slavic languages they are regular words, unlike toponyms.
There's no certain rule, and GG is assigned according to similar existing loanword nouns. For instance, most non-declining words borrowed from Romance languages are neuter due to their pronunciation (ending with -e or -o).

There are many exceptions, however:

  • RU, UA: whisky can be either masc or neutr;
  • RU: кофе (coffee) is masculine due to historical reasons: there was an older version кофий, ending with a consonant;

Naturally plural or paired items

Obviously, they are plural (grammatically, plural and dual are the same). Note that in most declensions, there is no difference between masculine+plural, feminine+plural, and neutral+plural.

  • Moccasin - shoes, naturally paired;
  • Jeans - naturally plural;

Noun groups

If one word has no declension while the other has, the GG of the whole group is given by the word that has declension.

  • Web-page, where Web has no declension while page is a common word (fem).


GG is given according to what category the noun belongs to. Look again at the rule above and think of "the city of New York".

  • Mississippi is a river; river is fem, hence, Mississippi is as well;
  • Ontario is a lake; lake is neutral;
  • San-Francisco is a city; city is masc;

There are lots of exceptions here:

  • Hudson, Rhein, although rivers (fem), are masc - probably, due to ending with plosive consonant;
  • Philadelphia, Nevada, are states (state is masc), is fem, again, probably, due to it ends with -a;
  • RU: Vienna is a city (must be masc), but is fem;
  • UA: city is neutral, but it does not apply on most city names; For example, Vienna is written Відень ['videnʲ], and it is masc;
  • Apache is tribe (must be neutral), but is plural;


GG is given by the keyword:

  • OUN stands for "Organization of United Nations", the highlighted keyword is fem (all words ending with -ion are fem)
  • USA stands for "United States of America", and it is plural;
  • TCP/IP is an "Internet protocol", masc;

And yes, if the loanword does not apparently belong to any of the groups above, it is often considered by its category or keyword.

Finally, if that was too boring, I have good news: there are lots of exceptions. :)

  • 1
    @Anixx: Right, if your language has such fixed rules for its native words, then it makes sense that they should be applied to foreign words as well. But don't people mess up the sex of abbreviations? They do here. They forget what the head noun was.
    – Cerberus
    Jan 8, 2013 at 16:01
  • 1
    @Cerberus some of the abbreviations become normal nouns over time. For example, ВУЗ (=Высшее Учебное Заведение) now can be written with small letters, "вуз" and has masculine gender even though the word "заведение" is neuter.
    – Anixx
    Jan 8, 2013 at 16:07
  • 1
    Russian and Ukrainian could be viewed as not being languages "with grammatical gender that has (almost) no morphological relation between the words and the genders(e.g. French)" (from the original question). Jan 8, 2013 at 21:33
  • 2
    @imz--IvanZakharyaschev You mean, there are any morphological reasons for applying GG? Well, then you have to explain what morphological reasons you find, which make aerosol feminine, but alcohol masculine in Russian. :) Jan 9, 2013 at 2:25
  • 1
    @imz--IvanZakharyaschev They are everywhere: toroid and spheroid (masculine) versus sinusoid and trapezoid (feminine). Jan 9, 2013 at 7:19

I have also been wondering about the question of gender-assignment in connection with Urdu, which I think is an outstanding example in what concerns your question. Technically it is New Indian, but the vocabulary is - apart from the basic Indic stratum - very very heavily influenced from mainly three sources:

  1. Arabic
  2. Persian
  3. English

Urdu itself distinguishes a (sort of default) masculine gender and a feminine gender. While Arabic is similar in respect to there being two genders - masc. and fem. - and the masculine being sort of a default gender, the other two sources of vocabulary do not distinguish gender.

Now this is just to give an idea, because as far as I know, loanword noun morphology in Urdu has not yet been studied extensively, the only available research I am aware of being a rather unhelpful dissertation (The Morphology of Loanwords in Urdu: the Persian, Arabic and English Strands).

By the very situation of three different sources with or without gender, most of the items in the list in aaa's answer can be found. Persian nouns ending in -ه, which used to be pronounced in Persian as a short final -a, are pronounced in Urdu with a long final -ā and thereby resemble a class of native masculine words in -ā. They are therefore almost invariably masculine. This would be point 3 in the above list:

If the borrowed word happens to have a suffix that the borrowing language uses as a gender marker, the suffix tends to dictate gender.

Similarly, Persian nouns with a final long -ī are classed as feminine in Urdu, since there is a class of femininae in -ī corresponding to the masculinae in -ā. These are not very frequent in Persian except from abstract nouns like هستی (existence) or نیستی (non-existence). One other example I found is تختی, sort of a diminutive, feminine in Urdu as well. The same thing happened to Arabic کرسی (chair), which is masculine in Arabic, but then feminine in Urdu.

An easy example for point 2 in the list

The borrowed word tends to take the gender of the native word it replaces.

would be the arabic word شمس (sun), which is feminine in Arabic, but masculine in Urdu, corresponding to the native words روی (रवि) and سوریه (सूर्य), the same is maybe true for Persian روز (day), which is masculine by default or via دن (दिन).

As to point 4

If the borrowed word rhymes with one or more native words, the latter tend to dictate gender.

it is hard to find an example which is for sure assigned a certain gender due to rhyming. But it may be noted, that even words which are phonemically identical like کان (Urdu: ear; Persian: mine) can differ in gender, the ear being masculine and the mine being feminine, in agreement with the predecessor of the native word, Skt. खानि, feminine. This indicates that at least in the case of Urdu, the "native gender" is stronger than the rhyming.

Unfortunately the coverage of English loans in the dictionaries I have is still very bad, since English loans, I guess, are still viewed as just a manner of speech that does not partake of "good Urdu", or rather "pure Urdu". But what is still interesting, is the handling of Arabic nouns, since Arabic has more or less the same system of genders.

Here we can note, that participles are by default masculine like in Arabic and the feminine forms are borrowed along with the masculine ones, e.g. عالم and عالمة or معشوق and معشوقة. But the verbal nouns do differ in gender, for example all the nouns from stem II of the form تفعیل are feminine in Urdu. This I guess, may be due to a general tendency of abstract nouns being feminine (maybe already in Sanskrit: suffixes -ता (f) and -त्व (neuter)).

It would be great, if someone had even more information and examples on that.


I think it's based on how the borrowed/new word sounds compared to existing words. For example, in French words ending in -age are typically male. But in German, the borrowed word "garage" became "die garage" - female. (E.g. die Frage, die Lage)

  • 2
    You might be partially right, but the example you selected to illustrate it is rather poor: the "g" in Garage is soft /ʒ/ [ɡaˈʀaːʒə], and hard in Frage [fʀaːɡə] and Lage [laːɡə]. It would work well in Portuguese though: embraiagem, garagem both feminine after viagem for instance . Jan 8, 2013 at 12:57

According to Ibrihim, 1973, p.61:

  1. If the noun is animate, natural gender tends to dictate grammatical gender.
  2. The borrowed word tends to take the gender of the native word it replaces.
  3. If the borrowed word happens to have a suffix that the borrowing language uses as a gender marker, the suffix tends to dictate gender.
  4. If the borrowed word rhymes with one or more native words, the latter tend to dictate gender.
  5. The default assignment is the borrowing language's unmarked gender.
  6. Rarely, the word retains the gender it had in the donor language.

Also, sometimes the choice is totally arbitrary, and may be different indifferent dialects. It may also change with time.

  • 3
    I'm seeing multiple papers from Ibrihim in 1973. Could you edit your question to say which paper you are citing? Maybe try to find a link on Google Scholar.
    – acattle
    May 12, 2013 at 7:35

In Swedish there are two grammatical genders, neutrum and utrum. Most nouns are categorized as utrum which is a merge of masculine and feminine genders in modern Swedish. New words are almost always utrum.

  • 1
    Björn, welcome to Linguistics! Usually we prefer longer and more elaborated answers. Please expand your answer by adding detail, context, examples, and backing up with references, this would increase your answer's quality. Poor and/or repetitive answers risk being down-voted. Pay a special attention to how the new loanwords get their grammatical gender in Swedish. "Almost always mostly utrum" does not seem to be that helpful for the original question. Jan 10, 2013 at 16:39
  • 1
    Easy counter-examples: "ett set", "ett disko", "ett aber". On the other hand, the list of neologisms from 2012 sprakradet.se/nyordslistor (PDF link on the page) has a handful of new nouns which are all uters, though apparently mainly so because they follow established patterns.
    – tripleee
    Jan 13, 2013 at 16:38

In Russian anything ending -а is feminine, anything ending -о and -е are neuter and anything ending with a consonant is masculine. Abbreviations case is determined based on the main noun in the abbreviation (this is the rule), but sometimes the speakers also apply the usual rules to the abbreviation as well, that is determine the gender based on the pronunciation, for instance "НАТО" ("NATO") is often used as neuter because of the -O ending, ВУЗ already became masculine this way. This happens mostly with familiar, often-used abbreviations which gradually become accepted as nouns.

The only remaining cases are the words ending with -и, -у whose gender is usually determined by the class to which the noun refers (city, river, organization).

There are multiple exceptions though.

  • 1
    Russian could be viewed as not being a language "with grammatical gender that has (almost) no morphological relation between the words and the genders(e.g. French)". Jan 8, 2013 at 21:30
  • @imz--IvanZakharyaschev, you probably should elaborate on that for us, the unenlightened.
    – theUg
    Jan 19, 2013 at 5:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.