.png){kind=logo}
.png){kind=logo}
Summary Advice
Gender is important in French, but it is particularly difficult for English people to acquire.
The simplest approach is to initially focus on a few rules:
The key one is:
If a noun ends with a silent 'e', then it is most likely feminine.
This involves knowing how a word is spelled, and how it is pronounced.
There are a few modifications to this, which are particularly reliable and useful
• Some nouns are inherently masculine or feminine.
• -ion is almost always feminine
• -té is almost always feminine
• -age endings are usually masculine.
use mnemonic image:
Marianne (female - liberté, in the station) + Mr Cheese (male, outside and fromage).
• foreign words are always masculine.
These will cover the greater majority of words. See below for how these rules were derived and a test and review of how effective they are - probably about 85%
The remaining words will develop with longer term use.
Use Gender Test to work on some of these.
Background
Gender is a key feature of French, as it can affect a number of other words – the main ones being le/la and un/une (articles), but also adjectives and verbs. It can also distinguish between different meanings (le livre = the book, la livre = the pound), and its effects can span across a number of words (ce sont deux étudiantes d'une grande ville qui se sont rencontrées au collège).
French gender has little to do with biological sex. It is more a grammatical marker that is used to link in with other words. It therefore provides a form of redundancy as it partially locks in the structure and meaning. For instance, in ‘la ------ était neuve’ you know that it can’t be referring to ‘(le) crayon’. The female gender (shown by 'la' and 'neuve') immediately shuts down half of all possibilities. Tucker (1967) found that most French people will therefore often say that the use of gender makes their language more precise.
In its earlier historical form, French did have regular gender markers at the end of words, with –a for feminine. Over time things evolved however, and it has become a very muted ‘uh’, or /ə/ sound, often referred to as a ‘shwa’. The final silent or muted 'e' at the end of nouns is therefore now typically feminine.
You can also see this trend in the gendering of adjectives and verbs, when an ‘e’ is added to correspond with a female gender word, as well as the feminisation of many nouns, to involve a final muted 'e'.
le chien (male dog), la chienne (female dog) - notice how adding the 'e' alters and emphasises the final consonant sound.
In English however we get by quite nicely without gender most of the time, apart from semantic gender – when there is some meaning behind assigning it. The obvious one being biological sex, but one can also see it in things like referring to boats as female – this is believed to emphasise their supportive, protective features. Interestingly, a recent study (Jakiela and Ozier, 2018) found that countries which make greater use of gender in their language are less equal in their treatment of women. As always though, it is difficult to know if the language is an effect or a cause of this.
English was also once a gendered language, but it lost this in the 13th century. Now, it seems that our brains don't develop the ability to automatically process this feature of language. The knowledge of gender is therefore one of the most difficult things for people learning French. Success with it is usually seen as quite an advanced ability.
Even fluent speakers of French as a second language will typically stand out by making occasional mistakes, which are apparetly quite jarring to native speakers. French people themselves make very few gender errors and appear to acquire this knowledge effortlessly, unconsciously, and from an early age. A study by Karmiloff-Smith (1979) of three year old French schoolchildren found that they were already able to operate at an 80% accuracy level, and were also able to apply their knowledge to unknown words.
Probably because of this, Tucker et al (1977) noted that French children’s grammar books did not include clues or explicit instruction for distinguishing gender. When it arises, children are just directed to see if using 'le' or 'la' feels right.
Where does gender knowledge come from?
One key aspect of developing gender knowledge is simply the pairing of a word with le or la. This is shown in a study by Dewaele (2015), who found that French speakers have a relative weakness with nouns which start with a vowel sound. Gender information is often missing for these words because of ellision. When this happens, the vowel for 'le' and 'la' are lost, and therefore no longer provide a cue to develop identification of the word's gender.
l'azote (masculine, despite the 'e')
However, the fact that Karmiloff-Smith found that young French children were able to guess the gender for unknown words shows that there is something in the actual word (sound structure and/or assocated knolwedge) which provides information about gender. To investigate this further, a study by Sokolik and Smith (1992) used a connectionist model which successfully developed gender identification based on the words alone. They also found that the trained network was then able to identify the gender of unknown words, with only a slightly reduced level of success. The neural network was evidently identifying a number of aspects to do this, but no single feature stood out.
The features of words which can be used to identify gender have been studied exhaustively, mainly focusing on word endings. There are a huge number of rules which have been generated, some of which are quite successful, such as the principle that –ion words are mainly female, and which has a 98% accuracy rate (Lyster 2005). Others are much less effective, but by combining a number of such rules, Lyster was able to achieve an overall gender identification success rate of about 80%. However, this is still nowhere near the level of native French speakers, and depends on a large number of different endings (about 400), which overlap and are easily confused.
Lyster did however establish that orthography (how a word is spelled) beats phonology (how a word is sounded) as a gender predictor. A rule of thumb which is commonly used, is that noun endings written with a final mute 'e' are associated with feminine gender. Of course, a criticism of this would be that the successful 3 yr olds mentioned above evidently were not yet able to use spelling. However, the silent e does have an effect on pronunciation (it emphasises the consonant sound as above for chien/chienne), and the children would also have had access to word meanings.
Given the success of Sokolik and Smith’s approach with a basic neural network, one can imagine that the natural learning process is based on a number of features, and that a lot of this is down to gender-based contexts. For instance, in whole phrases where there are a number of gender cues and agreements, the strongest ones being le/la. Tucker found that when native French speakers were not sure of a word, they mostly said they'd try it out with un/une or compare it with a similar word they already knew. Interestingly, only one in five said they would consider the meaning of the word.
Specific work by Lyster on direct teaching of gender knowledge has been shown to work, in terms of gender identification scores, oral activities, and text completion tasks. However, as always, the problem is to what extent this knowledge can be used spontaneously in real-life integrated contexts, how long-lasting it will be, and how motivating it is to focus on grammar, rather than meaning.
Developments
Some more recent work by Nelson (2015) has developed the idea of a partial semantic basis for gender in French words. His approach starts off by assuming masculinity (60% of words), then applying the further rules of spelling and sound, and then the contemporary meaning (semantics), as well as the earlier basis of the word. For instance, does it just have a suffix added (e.g. –ette, which means small, as in la camionette, and is typically assigned as feminine), or does the end constitute a key part of the word (e.g. –di, which means day, as in le Samedi, and has a masculine basis).
There are already a number of rules based on categories, such as the Sciences being Feminine (la physique, la chimie, etc), and the names of days of the week, month, and seasons being Masculine (le Samedi, le Mars, le Printemps). However, Nelson has extended this to cover a number of plausible aspects, such superordinate words (ones which cover wide general categories) being masculine, whereas words which cover more specific categories are femininine. So, for example 'le chemin' is a path, in the general sense of a way of getting to somewhere, but 'une allée' is much more specific, and refers to a path lined with trees, frequently in a garden or park, and which is generally rather short.
The really interesting thing is that by applying this rule system, Nelson arrives at an identification success level of up to 98%. This approach is therefore much closer to the level of native French speakers, than simply using the sounds and spellings of the ends of words.
One further aspect is that many of the masculine categories which he applies include more dominant features, such as size (le camion vs la camionette), and generality (un an vs une année). Along with other aspects such as the use of the masculine for commerce/finance, it is perhaps possible to see the operation of masculine dominance matched with gendered languages as described by Jakiela and Ozier( 2018).
Developing Gender Knowledge
Staying true to the theme of this site, I believe the learning of rules is not a good idea. The ability to apply knowlege which obeys rules is certainly the end point of fluent/native users of French, but this is normally developed and applied in an unconscious and automatic way. Although direct teaching of gender rules does improve scores on assessments, it is likely that such knowledge will be somewhat fragile and difficult to retain or to apply. It is also quite a painful exercise to engage in rote learning of lists of endings, only to find that there are then at least as many exceptions. I have personally learned quite a few such lists, but have never been able to retain or use them.
However, there are some broad principles which seem useful and require little effort to develop or apply.
Le chien La chienne
(btw, notice how feminising gives it a silent ‘e’ – see below)
This is of course means that you have to know the way the word is spelled. However, in general a final ‘e’ is silent, and you will hear the consonant sound that goes just before it (la mère). On the other hand, if there is a final consonant, this will usually be silent (apart from CaReFuL), and you will hear the vowel before it.
But, there are a few with a high level of accuracy, and it may be useful to focus on the most productive of these. French people also say that when they have difficultis with the gender of a word, they mainly try to think of a word that is spelled with a similar structure/ending that is already familiar to them (Tucker). I am therefore aiming at a single good exemplar word for a limited number of productive and accurate rules.
From the many examples in Lyster (2006), I have removed the rules which would already be covered by the principle of final silent ‘e’. As this is such a general approach, this leaves relatively few exceptions. I looked through these to identify those that would have the greatest impact, and ended up with just three:
Feminine
La station (-ion)
La liberté (-té)
Masculine
Le fromage (-age)
Look for some form of higher-level or encompassing category.
Higher, more general is masculine le chant (song)
Lower, more specific is feminine la chanson (more structured type of song)
Relative size
Bigger is masculine camion
Smaller is feminine camionette
Finance Masculine Le compte (overrides final e)
Foreign words are always masculine. Le selfie.
Most imported English words do not have an 'e' on the end, as shown on this this list.
However, 10% do, so this rule seems worth including, and the category is usuallyl self-evident.
Does it work?
I first generated a random list of 100 words, taken from the Collins French College Dictionary. I then simply applied all of the above rules, to see which were the most effective.
Best of all: The 'silent final 'e' rule' correctly identified 71% of words.
The '-ion rule' correctly identified 14% of words
The '-age rule' correctly identified 2%
The '-té' correctly identified 2%
The 'foreign word' rule identified another 2%
The 'assumption of masculinity' did not really seem to achieve much. As the above rules are quite clear-cut, there was no ambiguity to apply this to.
There was one word which I subsequently found could correspond to a semantic effect and would have reduced the error rate a bit. This was Azote (Nitrogen), which is a masculine noun, despite the 'e' ending. On checking further, it looks as though all of the other gaseous elements are masculine, despite their endings, such as Hydrogène. Another possible one is le crâne (skull), as Nelson (2005) considers that body appendages are feminine, and looking further, it's le torse (torso). There may have been other principles I missed, such as compound words and their derivations.
The majority of errors happened with masculine words which had an 'e' at the end. This is a typical finding in such studies.
However, adding up all of the correct words still gives a highly respectable value of 91%.
This is of course a somewhat dodgy analysis, in terms of sample size and the ratio of male to female words (there was a bias towards female).
BUT, looking around, these principles seem to be nothing new. A very similar set of advice can be found here, with findings that such rules apparently result in a coverage of about 85% in a range of situations. This therefore seems like a reasonable level to expect.
Following on, there are further groupings and developments which one could either pursue actively or just simply be allowed to develop by long term language experience. This would include the various categories proposed by Nelson, as well as those covered here.
References
Dewaele, J. (2015). Gender Errors in French Interlanguage: The Effect of Initial Consonant Versus Initial Vowel of the Head Noun. Arborescences, (5), 7–27. Link
Jakiela, P. and Ozier, O. (2018) Gendered Language. World Bank Group, Policy Research Working Paper 8464. Link
Karmiloff-Smith (1979) - quoted in Dewaele (2015) p. 10
Keller, R. (2013) La Barbe: Feminine Beards and Other Mysteries of French Grammatical Gender" Senior Honors Projects.Paper 326. Link
Lyster, R. (2006) French Language Studies 16 (2006), 69–92.
Nelson, D. (2005) French gender assignment revisited, Word, 56:1, 19-38. Link
Sokolik, M. and Smith, M. (1992) Assignment of gender to French nouns in primary and secondary language: a connectionist model. Second Language Research 8, 1; pp.39-58
Tucker, R. (1967) French Speakers' Skill with Grammatical Gender: An Example of Rule-Governed Behavior. PhD Thesis, McGill Univ. Link
Tucker, G. Richard, Wallace E. Lambert, and Andre Rigault (1977) The French Speaker's Skill with
Grammatical Gender: An Example of Rule-governed Behavior. The Hague: Mouton, 1977. Quoted in Keller (2013) P 4