Skip to navigation – Site map

The Conceptualization of Natural Gender in English

Elise Mignot
p. 39-61


Dans cet article je m’intéresse à la conceptualisation du genre naturel (les distinctions humain / non-humain et masculin / féminin), dans le but d’éclairer certains aspects du genre en tant que catégorie linguistique. Je propose que le genre est conceptualisé comme tenu pour acquis, toujours déjà connu, objectif, stable, non gradable, et que les inanimés sont conçus comme précédant les humains.

Il apparaît que ce que nous transférons dans le langage n’est pas le genre tel qu’il est dans le monde mais véritablement le concept de genre, ce dernier étant in fine déconnecté de son ancrage expérientiel.

L’identification de ce processus cognitif me permet de rendre compte du lien entre genre et anaphore, d’expliquer certains emplois du it non référentiel ; j’aborde également le lien entre genre et catégorie nominale, et enfin certains cas de non-concordance entre genre naturel et genre grammatical.

Top of page

Full text

0. Introduction

1My aim is to look at the conceptualization of natural gender in order to account for some aspects of gender as a linguistic category.

2A concept is taken to be a complex construct – complex in the sense that it consists of various elements, which are construed as inherently linked. The fact that the parts are inherently linked (rather than predicated of one another) means that this definition does not contradict the idea that concepts are non-compositional, i.e. neither true nor false (Panaccio 2011).

3I hypothesize that some concepts, e.g. that of gender, or at least some parts of them, are mapped onto language. That is, if we construe A as inherently linked with B (on the basis of the observation that in the real world A systematically co-occurs with, or involves, B), we may in some circumstances resort to A in order to express B, simply because A and B are construed as inherently linked. I will describe how gender as a linguistic category illustrates this cognitive process.

4Natural gender obviously concerns human beings (there are men and women, boys and girls), but not only; it also includes the animate vs. inanimate distinction, which is in fact often said to be the first gender distinction (e.g. Boone and Joly 2004: 208, amongst others). Although the masculine vs. feminine distinction applies to many entities in the world, such as animals, it is mostly relevant for human beings. Gardelle (2006: 270) points out that in present-day English animals are much more frequently denoted by it rather than by he or she. This points to a re-interpretation (from a human point of view) of the physical world: for us human beings, the relevant distinctions are 1) human vs. non human 2) within humans, masculine vs. feminine. These are the gender distinctions that I will consider in this paper.

5I will try and show that our experience provides a starting point from which we conceptualize natural gender. It is this conceptualization, i.e. the fact that A can be seen as inherently linked with B, and not the actual masculine vs. feminine, of human vs. non-human, distinctions, that may account for some uses or gender as a linguistic category.

6I will first note that gender is a denotatum which is not typically predicated of a referent. Gender is conceptualized as taken for granted.

1. Natural gender and predication

7In everyday language, natural gender is not talked about, not even mentioned. In other words, it is not the type of information that is predicated of a referent. We do not often utter such statements as: ‘it/she’s a woman’ or ‘my computer is a thing’. These sentences are perfectly grammatical of course, and they make sense, but they are not uttered. Or if they are, gender nous such as ‘woman’ or ‘man’ are not used to convey information about gender but some implicature, often close to stereotype: ‘She spends a fortune on perfume. – Well, what do you expect, she’s a woman’. In texts we find such utterances as ‘She was an extraordinary woman’ (where the noun is preceded by an adjective) but not *’She was a woman#’. In the first of these two clauses the noun ‘woman’ is only there to support the extra specification of the adjective (‘extraordinary’).

8In terms of information structure, this amounts to saying that natural gender is a denotatum that resists focalization. I assume, after Lambrecht (1994), that a canonical clause consists of a subject and a predicate, that the subject represents the topic, what is talked ABOUT, and that the predicate represents the focus. With the focus the speaker comments on the topic. (If I say: ‘I am hungry’, I am talking about myself, and about myself I add that I am hungry. ‘Be hungry’ is the denotatum that is focalized). It seems that natural gender is hardly ever focalized. It is not the kind of comment you make on a referent.

1.1 Data from a contextualised corpus

9To verify this, I have listed the nouns that denote natural gender, that is ‘man’, ‘woman’, ‘boy’, ‘girl’, ‘thing’, ‘people’, ‘person’ in a list of 2233 nouns extracted from a 12001 word-long contextualised corpus, consisting of excerpts from two novels (Coe 2007, Sillitoe 1959) and one screenplay (Allen 1983).

10Concerning the data extracted from Coe (2007), the first occurrence of a gender noun is on page 10 (the novel starts on page 1). Obviously humans have been mentioned before, in fact as early as page 1, but their gender is not explicitly mentioned.

11More generally, in the whole corpus there are only 63 gender nouns, i.e. 2.8% of nouns, a proportion that may seem low considering the importance of the gender criterion in our daily lives. Moreover, these nouns are never used in order to convey the information that someone is male or female, or that a referent is human or not. Not once do we find a statement such as: ‘he was a boy#’ or ‘it was a thing#’. When gender nouns appear in the predicate, i.e. as prdicative complements of a subject (13% of gender nouns), the gender of the referent is already known and the gender noun is only there to support additional specification. Examples are: ‘she wasn’t a delicate woman’ (Coe 2007: 46), ‘He was always a big man’ (Coe 2007: 47), ‘she was a different person’ (Coe 2007: 48), ‘He was a vigorous athletic boy’ (Coe 2007: 49), ‘She turned out to be a rather distant, unapproachable woman’ (Coe 2007: 52). Gender nouns may also be used in anaphoric contexts, in 32% of cases, e.g. ‘the girls had not come with them on that trip anyway’ (Coe 2007: 21), ‘the girls are fine’ (Coe 2007: 23), ‘the girls had left’ (Coe 2007: 25). Again gender nouns are not there to convey any information about natural gender. They are used as hypernyms; the gender information is supposed to be already known. In my corpus gender nouns never convey new information.

1.2 Introducing someone

12If natural gender was ever mentioned, that would presumably be when a new character is introduced. Let us take a look at two extracts from the novel where a new character is introduced.

  • 1 I underline the elements that are relevant for my analysis.

First, at the beginning of the novel, as is most commonly the case1:
When the telephone rang,
Gill was outside, raking the leaves into coppery piles, while her husband shovelled them on to a bonfire. It was a Sunday afternoon in late autumn. She ran into the kitchen when she heard its shrilling, and immediately felt the warmth of inside enfold her, not having realized, until then, how chilly the air had become. (Coe 2007: 1)

A person, Gill, is mentioned straightaway. Her gender is not mentioned, at least not explicitly. However the information is unambiguously conveyed, through 1) the proper name 2) the personal pronouns 3) the word ‘husband’ (if she has a husband, she’s a woman). Note in this respect that another character is introduced in these first lines of the novel, i.e. the husband. Again, it is not stated that he is a man, but the information is also unambiguously conveyed, this time through the lexical item ‘husband’, which includes the /+male/ semantic feature.

In the following extract, another character is introduced, who is first called a ‘stranger’.

Instead, Rosamond had divided her estate three ways: one-third to a stranger; a near stranger any way, as far as they were concerned. Her name was Imogen, and Gill had no idea where she was to be found these days, having met her only once, more than twenty years ago. (Coe 2007: 8)

13Again, the notion of gender is conveyed, but indirectly, via ‘her’ and the proper name ‘Imogen’. Note that this feels perfectly normal and cannot be ascribed to any stylistic effect, whereby something would be felt to be missing. The reader is not particularly surprised that natural gender is not predicated of the characters. In fact there is nothing typical of literature here, the same would hold for a real-life introduction: when we introduce someone, gender is not what we mention. We may say, for example: ‘This is Catherine, she’s my new neighbour’, but not *‘This is Catherine, she’s a woman’. Natural gender is of course relevant but is treated as taken for granted.

1.3 The embarrassment of mentioning natural gender

14In the rare cases where natural gender is not obvious, we are confronted with embarrassing situations. We all know that it is not socially acceptable to ask someone: ‘Are you a man or a woman?’ and we do everything we can to avoid this situation. In his discussion of pragmatic presupposition, Robert Stalnaker (1973) quoted in Lambrecht (1994) mentions a well-known type of conversational exchange. The situation is the following. A asks of B’s baby-girl: ‘how old is he?’ and B answers ‘She is ten months old’. The first speaker thinks that the baby is a boy when in fact it is a girl. Lambrecht, who borrows this example to Stalnaker and analyses it along the same lines, argues that when B replies ‘She’s ten months old’, she pretends that the fact that the baby is a girl is a shared presupposition, in order to re-establish a normal situation for a conversation. For Stalnaker this goes to confirm that a topic (normally expressed in subject position) is typically presupposed. I do not dispute that, but I would like to point out that this act of pretending would not take place in another situation. In this example, the speaker pretends that there is a common background because the sentence ‘Oh it’s a girl’ feels awkward. Let’s imagine a similar situation, where a wrong assumption is made about another type of referent (not a human being). If someone assumes that I recently moved in a house, when actually I moved in a flat, and if that person asks me ‘How big is your new house?’ I’m not likely to answer ‘My flat is rather big’. I would probably start with explicitly correcting the wrong assumption, and then I would move on. So I would say something like: ‘Oh it’s a flat actually. And yes, it’s quite big’. Robert Stalnaker’s example reveals not only that a subject is typically presupposed, but also that natural gender is not easily predicated of a referent. Gender is a denotatum that is conceptualized as taken for granted.

1.4 What are the denotata that are normally predicated of a referent?

15Now why don’t we predicate gender of a referent? What are the denotata that are typically predicated of a referent? I propose that we predicate e.g. what is subjective, i.e. what results from a personal perception, or interpretation of the referent. In ‘this is beautiful’, ‘this’ denotes something that the speaker and the hearer agree on, that they can both see, and the predicate expresses a personal judgement. A denotatum such as gender is not subjective and is therefore not likely to be predicated of a referent.

16What is predicated may also be (this is not mutually exclusive with what precedes) what the speaker and the hearer disagree on. For instance if someone says: ‘He’s young’, I can reply ‘He’s not so young, he looks younger than he is’. It is unusual to disagree about natural gender, precisely because it is an objective piece of information. Someone can be less young than he looks but not less ‘man’ than he looks.

17We also predicate something that we want the hearer to know or to remember. For example, if I say ‘I am hungry’, I want the hearer to know that I am hungry, so that we may act upon it. Gender is not likely to be predicated because it is supposed to be already known.
There are however a few counter-examples to the tendency I’ve just described.

1.5 Some cases where gender is predicated of a referent

18The reason for the fact that natural gender resists focalization cannot be that gender is not relevant. It is of course extremely relevant in our daily lives, in terms of social interaction, reproduction, survival of the species... It is in fact so relevant that it is the first predication to be made about a person: when a baby is born one says: ‘It’s a boy’ or ‘It’s a girl’. The next stage is that the baby is given a (first) name, which very often (although not always) includes the /+male/ or /+female/ semantic feature – Then natural gender is not supposed to be discussed, or even mentioned any more, at least, by adults. It is taken for granted.

  • 2 This remark is based on personal observations. It is very difficult to test empirically in large el (...)
  • 3 My translation from French of personally collected data (sentences uttered by four-year-olds).

19It has to be noted, however, that one may find gender predicated of a person in children’s conversations2: ‘I can’t wear those clothes, I’m a boy / girl’; ‘he used a pink pencil, he’s a girl’3. Adults, however, do not seem to have these conversations. And even when children do, this does not mean that children are not aware of their gender. With ‘He’s a girl’ the child means that he knows perfectly well that his/her interlocutor is a boy, but that in his (stereotyped) opinion he behaves like a girl. Or when a girl says ‘I’m a girl’ she knows that her interlocutor knows that, she is only justifying a refusal to wear something. It is not a ‘first’ predication; in other words gender is not supposed to be new information.

20What we may conclude so far from these data is that gender is conceptualized as taken for granted. Now the reason for this is probably the very nature of gender in real life. It is an attribute that does not change throughout life, contrary to the other attributes that can be predicated of a person. Take again the birth situation: of the three attributes that are predicated of a baby (gender, height and weight), gender is the only one that does not change throughout life (except of course in exceptional circumstances, which are probably thought of as too marginal to be integrated in the conceptualization of gender).

21I will now note a convergence between the above-mentioned discourse phenomena and the structure of the lexicon. Both levels illustrate the fact that gender is conceptualized as taken for granted.

2. From pragmatics to lexical semantics

22The structure of the lexicon offers other indications that natural gender is conceptualized as ‘taken for granted’.

2.1 The missing nouns

23The English lexicon seems to be organized in such a way that there is no noun which means ‘male human being’ or ‘female human being’, that is that holds for a male or female human being throughout his or her life. The noun ‘woman’, for instance, denotes a female human animate, but further categorizes the person as an adult. Gender nouns such as ‘man’, ‘woman’, ‘boy’ or ‘girl’ all add to the gender specification an extra semantic feature. In other words, where human beings are concerned, there is no noun which expresses natural gender only. I can only conjecture that the reason for this is that gender is construed as such a basic specification that it is always back-grounded in order for the noun to convey another feature of the referent. This is another indication that gender is construed as ‘taken for granted’.

24Dictionary definitions of gender nouns make this phenomenon clear. In the Shorter Oxford English Dictionary the first definition of ‘woman’ is: ‘An adult female person, as opposed to a man or girl or both’ (my emphasis). The ‘adult’ feature is also present in the other senses of the word listed in the dictionary, e.g. ‘servant’ or ‘lover’, even if it remains implicit in the definition. The only meaning of that word that may seem to contradict our view is the third one, where the noun is used in its generic sense: ‘Without article. The female human person, esp. viewed as a type; the female sex.’ However I would argue that even there ‘woman’ does not denote all human feminine beings but only adults, as suggested by the two examples which illustrate that meaning: ‘Dryden Woman’s a various and changeful Thing! S. Richardson Woman is the glory of all created existence: – but you, madam, are more than woman!’. In both examples the noun ‘woman’ occurs in the context of (love) relationships between men and woman, i.e. adult human beings. Note in this respect the form of address ‘Madam’ in the second example, which includes the /+adult/ semantic feature and suggests that when the speaker uses the noun ‘woman’ he is thinking of adult female human beings, rather than of all female human beings.

25The same observations hold for the noun ‘man’. When used to refer to a male person (I will come back to its other use) it categorizes the person as a male human being but also contains the /+adult/ semantic feature, as the definition of the Shorter Oxford English Dictionary makes clear: ‘an adult male person, as opposed to a woman or boy or both’. The nouns ‘boy’ and ‘girl’, conversely, include a /-adult/ semantic feature, and therefore do not denote natural gender only either.

26The nouns ‘male’ and ‘female’ do not qualify to fill the gap in the lexicon, as 1) they are derived from adjectives; 2) they do not only denote human beings but also animals; 3) they are rarely used for humans anyway (we do not normally say: ‘my neighbour is a female of extraordinary beauty’ or ‘a female came round to see you’).

27This hole in the lexicon seems quite extraordinary, considering the relevance of the masculine vs. feminine distinction in our daily lives and our otherwise prevailing anthropocentrism: we mainly speak about ourselves, we locate things and events relative to ourselves – think of Benveniste’s expression ‘l’homme dans la langue’ (1974: 223-285) – but we have, as it were, no nouns for ourselves, or at least for ourselves as classified according to the most basic criterion of categorization of our species. These remarks on English about the structure of the lexicon seem to have some cross-linguistic validity; I have data going in the same direction for Arabic, Finnish, French, German, Greek, Inuit, Japanese, Polish, Portuguese, Russian, Spanish, Swedish.

28The data concerning Inuit is all the more interesting as this culture may be thought to contradict the view taken here that gender is conceptualized as stable (it does not change throughout life) and objective (gender assignment does not result from any individual appreciation). Up to fairly recent times, some Inuit new-borns were assigned a social gender different from their natural gender; a boy could thus be brought up as a girl before re-adjusting to his ‘real’ (i.e. natural) gender at the age of puberty (Saladin d’Anglure 1977 and 2010, Héritier 1996: 202-203). Although from this description it looks as if gender is not held as a stable datum, in fact it should be pointed out that the assignment of the social gender is not due to any personal decision; the shaman who states the social gender is supposed to only state a fact, namely that the new-born’s soul (independently of the new-born’s appearance) is male or female. Afterwards that social gender does not change throughout life, even when teenagers re-adjust to their natural gender. Moreover, the fact that individuals eventually re-adjust to their natural gender indicates that even in that culture natural gender is construed as a non-negotiable, basic fact of life that can only be accepted as such, and is therefore not likely to be discussed.

29The structure of the lexicon at any rate is such that, just like in English, there is no ‘pure’ gender noun. My interpretation of this fact is that a person is always thought as pre-categorized (on the criterion of gender). This entails that there is no pure gender noun; gender is only one (back-grounded) semantic feature.

  • 4 Mummy and I are both... hmm... women... well... no… I’m not a woman. I’m a girl and Mummy’s a woman (...)

30We adults do not even realize there is this hole in the lexicon. We make do and find a way round if we need to refer to male or female human beings irrespective of their age. For instance, if an audience includes adults and children, we may hear: ‘ladies and gentleman, boys and girls’; the speaker addresses the older and younger members of the audience with different words. Children, on the other hand, sometimes look for the missing noun. I have observed it several times (in French), for example in the following utterance, produced by a child who was looking for a noun which may subsume ‘girl’ and ‘woman’ in order to describe what she and her mother both are: ‘maman et moi on est des ... heu... des... dames, heu... non moi je suis pas une dame heu ben moi je suis une fille et maman est une dame’4 (unelicited data from a 4.5-year-old). One may reasonably hypothesize that similar utterances could be produced in English, as the corresponding nouns exhibit the same semantic features (/+adult/ for ‘woman’ and /-adult/ for ‘girl’). It appears that at one point children, or at least some children, need to express natural gender, then logically assume that there’s a word for it, and cannot find that word.

31Now we may wonder if there is a noun which may denote all human beings, male and female. If not, this means that there are more than the two previously identified ‘missing nouns’ in the lexicon.

32Possible candidates are ‘person’ (and ‘people’), ‘human being’, and ‘man’ in its generic sense; but all these expressions are somewhat restricted in their usage and do not really qualify to fill in the gap for the “human being” meaning, for the following reasons. ‘Man’, if it refers to a human being of either sex, cannot be used in a specific context: ‘there is a man standing over there’ is not synonymous with ‘there is someone standing there’. ‘Human being’ is compatible with specific reference but it is highly unusual to refer to someone with that expression: ?’there is a human being at the door’ or ?’a human being came round to see you’. Another restriction concerning ‘human being’ is that it also sounds unusual in anaphoric contexts: ‘A new neighbour moved in. The human being seems very nice.’

33The noun ‘person’ seems the best candidate, as one may say: ‘there is a person standing at the door’ (although ‘a person came round to see you’ seems less likely than ‘someone came round to see you’). Its usage is however also restricted, as it does not work very well in anaphoric contexts either: ‘?the new neighbour came round to see you. The person seemed very kind’. Another restriction for the noun ‘person’ is that, although in theory it applies to all human beings, in practice it would be unusual to use it for a baby (?’Anna’s baby-brother is a charming person.’).

34This lack of ‘pure’ gender nouns is not restricted to human beings. There is no noun that means ‘non-human entity’ (i.e. that would subsume animals, plants, concrete objects and abstract ideas), even though, as I have observed at the beginning of this paper, the first gender distinction is that which opposes humans and non-humans.

35It can even be argued that there is no noun that means ‘inanimate entity’. ‘Thing’ is of course the best candidate, but it does not hold for all inanimates, as it does not work for abstract ideas, such as ‘courage’ (one may say ‘Courage is a wonderful thing’ but not *‘His courage was very much admired. The thing was talked about’; in other words, ‘thing’ is not appropriate in anaphoric contexts. ‘Thing’ does not even work for all concrete referents, only for prototypical ones (i.e. material, bounded entities). In particular it cannot be used to designate places. Even though ‘London’, for example, includes the /+inanimate/ semantic feature we cannot say: *‘London is a nice thing’, the correct alternative being ‘London is a nice place’.

36Finally, I observe that there is no noun that subsumes human and non-human referents. The best candidate would be ‘entity’, but that is a rather technical term, not used in everyday language; it is not a ‘basic level term’, in the sense of Rosch (1976 and 1978). Again this means that each time we name, we have already classified the referent. Natural gender is supposed to be old, backgrounded information.

37I therefore observe a convergence between what happens at the level of pragmatics (we do not predicate, i.e. focalize, natural gender) and what happens at the level of lexical semantics (there is no ‘pure’ gender noun). In all cases it appears that gender is not a denotatum that is foregrounded; gender is construed as taken for granted.

2.2. The missing pronouns

38Yet another sign that gender is construed as always taken for granted is that there is no way to ask what something is if we do not even know whether we are talking of a thing or a human. To make things clearer, let us imagine the following situation. I stumble on something in the dark, which could either be a thing or a person, and I want to identify that something. If I ask a question, I have to choose between ‘What is it?’ and ‘Who is it?’, i.e. I have to pre-suppose a basic categorization of the referent, even though, paradoxically, I am in a context where I seek identification. There seems to be no way to phrase what I want to phrase in this situation. In the same way that there are missing nouns there is a missing interrogative pronoun.

39Similarly, there is no pro-form that subsumes ‘something’ and ‘somebody’. If I make out a shape in the distance and do not know whether it is a thing or a person, I have no way of saying ‘I see something or somebody’. There is no pronoun that cumulates (or neutralizes) the /+human/ and /-human/ semantic features. The pronoun must select a feature, which is that of the category.

40The same applies to ‘anyone’ / ‘anybody’ and ‘anything’. There is no term that can express the idea that I cannot see anyone or anything. The referent has to be previously categorized.

41This again shows that natural gender (here taken as the human vs. non-human distinction) has to be taken for granted. Gender is construed as being always already known, a referent is supposed to have been previously categorized.

2.4. Summary

42After these incursions into the domains of pragmatics and lexical semantics, let us now sum up what makes the conceptualization of natural gender, i.e. what we see as inherently linked with natural gender. The following list is not necessarily exhaustive. Natural gender involves a two-term opposition (either masculine vs. feminine or human vs. not human); it is an objective piece of information; it is not supposed to be predicated of a referent; it is taken for granted (or, in other words: it is supposed to be always already known); it is stable (i.e. it does not change), this last characteristic being the reason why it is taken for granted. One may also add that natural gender is not gradable – I will come back to this characteristic. Note that what constitutes the concept may be a simplified version of what we experiment in our daily lives. Although in real life gender may, in some rare cases, change (some individuals may switch from male to female or vice-versa), the most usual and typical (i.e. entrenched) situation is that gender does not change (there is no switching between human and not human), and that is what is selected in the concept.

43I will now address the question of the link between natural and grammatical gender. How can the conceptualization of natural gender account for some aspects of grammatical gender?

3. From natural to grammatical gender

44In English, grammatical gender only appears on pronouns, not nouns, as used to be the case and as is the case in many languages (e.g. French). The English pronouns concerned with grammatical gender are third person personal pronouns (he, she, it) as well as interrogative and relative pronouns (e.g. who vs. which). The last two types only illustrate the human vs. non-human distinction, whereas personal pronouns represent the only area in grammar where the three genders are coded. Here I explain how the conceptualization of natural gender may help us account for some uses of third person personal pronouns.

3.1. Gender and anaphora

45Third person personal pronouns exhibit grammatical gender (Huddleston and Pullum, 2002: 1468), which in most cases (see Gardelle 2006 and the last part of this paper for counter-examples and discussion) reflects the natural gender of the referent. One may add that in the case of third person personal pronouns, gender is the only piece of information which is given. This is not true of interrogative and relative pronouns, which are marked for gender but, with the ‘wh’- sub-morpheme, also convey the notion that some information is missing (Rotgé and Lapaire 2004: 253).

46Personal pronouns are also, at least in most of their uses, anaphoric (see statistics about the uses of ‘it’ in Biber (1999) and a critical review of Biber’s figures in Labrosse (2003)), so that one may suppose a conceptual link between gender and anaphora. Indeed, as personal pronouns signal anaphora without the help of any additional morpheme or sub-morpheme, it may be conjectured that what signals anaphora is precisely gender.

47Third person pronouns signal that the referent has already been mentioned, or is supposed to be known to the speaker and the listener. This is what distinguishes them from first and second person pronouns, which are deictic (Huddleston 2002: 1468). Halliday (1976: 48) writes that ‘Only the third person is inherently cohesive, in that a third person form typically refers anaphorically to a preceding item in the text’. He also writes that ‘a third person form is assumed to be anaphoric unless the context or situation makes it quite unambiguous.’ (1976: 51). Commenting more specifically on ‘it’ (not on all personal pronouns), Cotte (1996: 203) points out that not only is ‘it’ anaphoric but, one may add, very anaphoric. ‘It’ signals a second anaphora. Cotte notes that third person pronouns are often preceded by another, more explicit, form of anaphora, such as a definite noun phrase, which includes either a definite article or a demonstrative determiner. A referent is denoted by ‘it’ if it is very stable, so well established in the speaker’s and hearer’s minds that there is no need for a lexical item to identify the referent. The author analyses such sentences as ‘that’s it’, or ‘this is it’ along the following lines: the speaker establishes a relation between something recently identified (which is conveyed through ‘that’ – e.g. what you’ve just said or done, what I can see at the time of speaking), and an older, previously established referent (‘it’) (Cotte 1996: 209).

48Now, if third person personal pronouns are (often) anaphoric, and if they encode gender, one may wonder what the nature of the relationship between gender and anaphora is. In other words: through what conceptual link is grammatical gender apt to signify anaphora?

49Grammarians do not even seem to formulate the question. The link between gender and anaphora is (at best) noted but not accounted for. Halliday (1976) for example establishes no conceptual link between anaphora and gender, nor does Corblin (1995), who provides a detailed study of anaphora. One could indeed think of other good candidates to express anaphora, such as the lexical items ‘continue’, ‘back’, ‘stable’, ‘previous’… Why resort to gender? What is the conceptual link between gender and anaphora? Cognitive linguistics, which did investigate the link between natural gender and language, is of no more help. Lakoff (1987) after Dixon (1982: 185-205) on Diyrbal established a link between gender and categorization (noun classes) (see also Dixon 2002). But no link is made between gender and anaphora, even though gender is not just exploited to signal categorization. Langacker (1991 2: 491) uses the notion of ‘common ground’ and Van Hoek (2007: 895) analyses anaphora in terms of accessibility and conceptual distance. Again the link between gender and anaphora is not accounted for.

50If the only information coded in pronouns is gender, then it is likely that there is a (semantic) affinity between gender and anaphora. I propose that this affinity may be due to the fact that natural gender is conceptualized as taken for granted (it is not supposed to be new information). Isn’t this exactly what anaphora is about? Anaphoric pronouns signal that a referent is already known, sometimes (but not always) because it has already been mentioned, and does not need to be foregrounded any more. What is activated here is not the whole concept of natural gender, but the ‘already known’ part of the concept. If we conceptualize gender as (amongst other things) already known, we may then use gender to (indirectly) signal that something is already known.

3.2. Some cases of non referential ‘it’

51Some uses of ‘it’ are not as clearly anaphoric. The ‘it’ of weather verbs, for instance (‘it’s raining’, ‘it’s hot’, ‘it’s sunny’, ‘it’s windy’, ‘it’s snowing’), does not seem to stand for any previously identified referent. This ‘it’ is often described (e.g. Huddleston and Pullum 2002: 1482) as being a non-referential, dummy subject, only there to fill in the slot of the syntactic subject. It is said not to represent a semantic argument. This also holds for sentences uttered to indicate time: ‘it’s ten o’clock’.

52Even if these pronouns do not clearly indicate anaphora, the question of their relationship with gender still stands. Weather verbs cannot take any other subject than ‘it’ (we do not say *‘the weather is raining’, *‘the sky is raining’, *‘the situation is raining’, *‘the place is raining’, or *‘the rain is raining’), which means that gender, for whatever it stands, is the only possible option for a subject.

53It has been proposed (Bolinger 1977, Cotte 1996 : 211) that the pronoun ‘it’ means that the place where it is raining exists independently of the weather, and, more importantly, precedes it: ‘Quand it s’emploie pour dire le temps chronologique ou météorologique il n’est pas anaphorique et il semble ne correspondre à aucun référent identifiable. Mais le choix de it n’est pas fortuit. Le temps des horloges et le temps qu’il fait ne naît pas avec l’énonciation mais la précède. C’est un donné indépendant dont l’énonciateur prend acte’ (Cotte 1996 : 211). Cotte’s interpretation ties in with what I have suggested so far. So does Neveux’s analysis of adjectival ‘it’ in this volume. If natural gender is conceptualized as pre-dating any other specification of a referent, grammatical gender may be used to signal that something is taken for granted, in the sense that it precedes any other specification.

54To this interpretation I would like to add that what is most relevant here in the conceptualization of natural gender is perhaps not only the fact that it is taken for granted, but also the reason why it is taken for granted, i.e. its stability. (As seen before, in real life natural gender does not change, contrary to other attributes like height or weight.) Let us note indeed that in our examples (‘it’s raining’, ‘it’s hailing’ etc.) the predicates describe extremely variable states of affairs (the weather, the time of the day). I hypothesize that the subject of these sentences (‘it’) represents stability and contrasts with the variable character of the predicate.

55In this perspective it is worth bearing in mind that subjects are synthetic, whereas predicates are analytic. In a typical sentence such as ‘he is hungry’, the personal pronoun designates a person independently of the processes in which he or she participates. Predicates, on the contrary, mention one thing (amongst many) about the subject referent. In this respect weather sentences seem to perfectly fit the pattern. The pronoun ‘it’ refers to a place in terms of its unity and permanence, i.e. independently of the weather, whereas the predicate denotes one (temporary) aspect of that place. It probably makes all the more sense to formulate things this way as the weather may affect quite strongly the appearance of a place: one and the same place may look like two different ones when it is covered in snow or scorched by the sun. It is probably all the more important, for this reason, to stress the permanence and unity of the place, beyond incidental weather conditions. This is what is achieved, although indirectly, by the use of a pronoun which only expresses gender.

56Note that if interpreted in this way, ‘it’ is a very typical subject, and that there is no need to treat weather expressions as thetic sentences where the whole clause, rather than just the predicate, is in focus (Lambrecht 1994). In ‘it is raining’ it is really the predicate that is focalized, in contrast to the subject, which is seen as denoting the place independently of variable weather conditions.

57Contrary to the ‘it’ of weather verbs, in ‘it’s ten o’clock’ the pronoun may alternate with ‘the time’: ‘the time is now 10 o’clock’, but the ‘it’ expression is the most common, and the same analysis applies. The type of predicate resembles the preceding one in that it is highly variable (the time of the day keeps changing). I therefore interpret this ‘it’ as signifying stability, in contrast to the variability of time. The point of ‘it’ is not so much to be referential as to provide a stable background to the variability of time.

58One last point to be resolved about weather expressions is the use of the neutral (rather than masculine or feminine) gender. If gender is meant to signify that the place precedes the weather conditions and exists independently of them, either one of the three genders could do. However we normally say ‘it’s raining’ rather than ‘he’s raining’ or ‘she’s raining’ (I will come back to possible exceptions to the standard formulation). Why is it the neutral gender that is selected? A place is a non-human referent, and an inanimate. This could be a simple explanation: we normally refer to inanimates with ‘it’. However I would like to suggest another one (which is not incompatible with the first one). I propose that we construe inanimates as pre-dating humans: when we were born, the place where we were born and what was in that place (i.e. inanimates) existed before ourselves. It may then be the case that ‘it’, the neutral pronoun which typically stands for inanimates, signifies here ‘what was there before’, i.e., the place before, or, independently of, weather conditions. Here this ‘before’ is understood as not just chronological but also logical, or even ontological.
The way gender is conceptualized may help us account for another phenomenon, i.e. the link between gender and categorization.

4. Gender and categorization

59As often pointed out (e.g. Taylor 2003), categorization is essential to human beings; it is the means through which we make sense of experience, and this cognitive process is reflected in language.

60The link between gender and categorization has been repeatedly established, particularly from a cross-linguistic perspective, through the study of the various systems of nominal classification (noun classes and classifiers) (see Senft 2007 for an overview of the literature). It is generally held that gender systems are a sub-category of noun classes. Hagège (2001: 78) proposes that in languages such as French or German, gender is a special, highly grammaticalized case of noun classes; Corbett (1991: 5) takes agreement as the defining characteristic of gender and states that many systems of nominal classification therefore fall within his cross-linguistic study of gender; Dixon (1986: 105-107) holds the view that languages with gender obligatorily classify all their nouns into formal classes and that gender systems are the most limited systems of nominal classification with respect to the number of their classes.

61In English, however, nouns do not exhibit grammatical gender, which may be the reason why the link between gender and categorization, extensively researched for other languages, is not dealt with. In the following section I try and show that the link between gender and categorization does apply to English. One indication of this is to be found in the structure of the lexicon.

4.1. The structure of the noun lexicon

62English nouns are concerned with gender (even though not with grammatical gender), in the sense that they include in their definition a gender semantic feature (/-human/ or /+human/, and, in some cases, /+male/ or /+ female/). A noun therefore classifies a referent as either human or not human; the relation between gender and categorization applies to English in the sense that gender is a structuring principle of the lexicon. This is a crucial characteristic of the nominal category as it holds for nouns only (not for any other part of speech). Adjectives, prepositions, adverbs, etc. do not include such a feature. Some adjectives, such as ‘blonde’ may occur preferentially with animate nouns but, in the case of ‘blonde’ as in that of all adjectives it cannot really be said that /+human/ is a semantic feature of ‘blonde’ itself but of the noun it is predicated of. In the same way, some verbs, such as ‘walk’, are more likely to be predicated of a human referent, but cannot be said to include the /+animate/ semantic feature.
Another reason why this is a crucial characteristic of nouns is that it holds for virtually
all nouns.

63For some nouns, that gender feature can be morphologically coded as a suffix, e.g. ‘-ess’ for the feminine in ‘actress’, but this should not obscure the fact that simple nouns also include a gender semantic feature. Words such as ‘book’, ‘chair’, ‘table’, ‘story’, ‘bottle’, ‘envelope’, ‘paper’, ‘card’, ‘magazine’, ‘picture’ or ‘chair’ unambiguously denote inanimate referents. ‘King’, ‘woman’, ‘neighbour’, ‘doctor’, ‘nurse’, ‘teacher’, ‘husband’, on the contrary, unambiguously denote human animates. This is quite a remarkable fact as there is no other feature that can be said to be systematically used in the lexicon. The feature /+furniture/, for example, which is part of the definition of ‘table’, cannot be said to be a categorizing principle applying throughout the lexicon, as the semantic features of other items such as e.g. ‘flower’ or ‘pen’ do not include /- furniture /.

64In addition, the semantic feature of gender seems to be amongst the very few that are easy to describe. In fact it is now established that definitions in terms of semantic features (such as the transformational-generative approach of Katz and Fodor (1963), Katz and Postal (1964), or the ‘componential analysis’ of Leech (1981)) are at least partially inadequate (Taylor 2003: 27-39), to the extent that this was one of the reasons for the development of alternative approaches such as the prototype theory (Rosch 1978).

65Contrary to what is sometimes said, then, there are few epicenes in English, if the primary gender criterion is taken to be the human vs. not human distinction. Note that what holds for common nouns also holds for proper nouns: ‘Manhattan’ denotes a place and includes the /- human/ semantic feature, while ‘John’ denotes a person and conveys the /+human/ semantic feature. Most first names include an additional /+masculine/ or /+feminine/ semantic feature, but if they do not, they at least always include a /+human/ semantic feature.

66The absence of epicenes is even more striking if we take into consideration the fact that the meaning of many nouns includes a locative relation, a process, or the attribution of a characteristic that could potentially apply to humans as well as non-humans. The noun ‘neighbour’, for instance, can only be applied to a human being. If an object is located near another one, it cannot be called a ‘neighbour’.

67Nouns that denote an entity in relation to a process differ when they denote animates or inanimates. For example, ‘a view’ denotes an inanimate. A person can be looked at, but is not called a view. In the same way, a person can be used for some purpose, but it cannot be called a device; ‘a need’ is something that is needed, but not a person who is needed; ‘relics’, ‘remains’, ‘leftovers’ denote inanimates only, although people may stay behind or be forgotten; ‘a surprise’ is a concrete object, or an event, but not a person (if someone arrives unexpectedly you cannot say that he or she is a surprise); the default meaning of ‘group’ is ‘group of people’ (if I group books together I’m not likely to call them a group); ‘a couple’ denotes two people (if I have two books in my bag I do not call them ‘a couple’- of course I may say that I have a couple of books, but in that expression ‘couple’ is part of a complex determiner); ‘news’ denotes an inanimate (a person that has just arrived is not ‘news’); a person cannot be called ‘the end’ or ‘a side’ even if he or she is at the end of a line, or at the side of a group; a relative is a person, not, say, a book that is related in some way to another, or a printer that is connected to a computer); a movie is an inanimate (if a person moves, he or she isn’t a movie); ‘the 1970s’ denotes a period of time, not people born in 1970; an annexe is an inanimate (if a person is added to a team, it is not called an annexe); although both doctors and drugs (or medicine) cure people, the fact that we have two different words (‘drug’ and ‘doctor’) reflect the relevance of the human vs. non human distinction.

68There are of course some counter-examples to this distribution of the lexicon across gender lines. ‘Companion’ or ‘reader’ for instance, may denote a person or a (reference) book. About ‘reader’ it can be noted however that ‘read’ is not used in the same way when the noun denotes a book or a person: a person reads but a book is read. This different orientation (active vs. passive) may account for the two uses. A similar remark can be made about ‘diner’ (a person who dines or a place where people dine / eat). Another apparent counter-example is ‘eccentric’, which may denote a person or an inanimate (an eccentric, i.e. not central, hub). But in that case the meanings are quite different: when the noun denotes an inanimate its meaning is literal (or: compositional), i.e. ‘away from the center’, whereas when it denotes a human it is not. This example therefore supports my hypothesis that it is difficult for one and the same meaning to be distributed across the gender line.

69Other counter-examples to the tendency for nouns to either denote an animate or an inanimate are to be found in nouns such as ‘party’: a political party may be seen as a group of people or as an abstract (inanimate) entity. I suggest however that even when they refer to groups of people these nouns actually denote inanimates, as the representation of a group involves the representation of an entity ‘above’ the individuals that constitute the group, i.e. an objectification. The unity of the group is more important than the addition of the people that constitute it, as illustrated by the possibility of singular agreement; if one adds humans together the result is of course a collection of humans but the new unit is probably of another kind. Other examples are ‘meeting’ which can either denote an inanimate (the fact that people meet) or the group of people that meet, ‘household’ which may denote the people that form a household but can also be understood as referring to a more abstract entity, or ‘government’. A noun like ‘top’, which normally designates an inanimate but can also be used in a phrase such as ‘the top of the company’ (i.e. the leaders), illustrates the same phenomenon, as it can only designate humans if it designates several humans (the default interpretation of ‘the top of the company’ is ‘the group of people that lead the company’, not ‘the managing director of the company’).

70This tendency for nouns to designate either a human or a non-human referent (not both) is particularly striking when one takes a look at derived nouns, such as nouns ending with the ‘-er’ suffix. Although they can potentially denote both types of entities (people or machines, typically), some are restricted to humans even though in terms of meaning potential they may also denote inanimates. For example, a hairdresser is a person, not a comb; a hairdryer is a thing, not the person who dries your hair at the hairdresser’s, a baker is a person, not an oven. Again, just like for simple words, I notice that for one and the same process we have different lexicalizations (‘baker’ vs. ‘oven’, ‘hairdresser’ vs. ‘comb’, etc.) depending on the gender of the participant who (or which) bakes / combs, etc.

71Others, such as ‘record player’, ‘cassette recorder’, ‘computer’, ‘calculator’ are restricted to non-humans. Someone whose role is to calculate (or compute) is not a calculator or a computer, but e.g. a mathematician. A typewriter is a machine, not a person who types; a scanner is also a machine, not the person who operates the machine. ‘Word-processor’ denotes an inanimate, even though when I write I process words.

72These remarks hold for deadjectival ‘-er’ nouns: a stranger is someone I don’t know but not something I don’t know. (I can say ‘this man is a stranger to me’ but not *‘this book is a stranger to me’). A foreigner is a person, not e.g. a foreign habit.

73This phenomenon is admittedly not true of all ‘-er’ nouns, i.e. some may denote humans and non-humans. For instance ‘smoker’ may denote a person (who smokes) or a train carriage where it is allowed to smoke. But the second meaning is only likely to occur in a specific context, where train carriages are differentiated according to their function (‘diner’, ‘sleeper’, etc.), which amounts to saying that the ‘person’ meaning seems to be the default one (the one that first springs to mind in the absence of a particular context). I therefore observe a fairly big discrepancy between what is possible in theory and what actually happens in language, which again points to a specialization of the lexicon along gender lines.

74It appears that gender is relevant in English, and that it has to do with categorization. As mentioned before this link between gender and categorization is only relevant for the nominal category, not for any other parts of speech.

4.2. Gender and the nominal category

75In English, gender concerns the nominal category, i.e. nouns but also pronouns (see Langacker (1987: 242) for reasons to posit that pronouns are a sub-part of the nominal category). Nouns are concerned with semantic gender; this is what I have described in the preceding section. Some pronouns, on the other hand, exhibit grammatical gender. That is the case for third person singular personal pronouns (‘he’, ‘she’ and ‘it’), interrogative pronouns (e.g. ‘who is here?’ vs. ‘what is here?’) and relative pronouns (e.g. ‘the person who came’ vs. ‘the book which is on the table’).

76About interrogative and relative pronouns, note that it is only the pronominal category that is concerned with gender. Determiners are not gender sensitive, as appears in the following expressions: ‘which book would you like to read?’, ‘which person would you like to see?’, ‘the person whose book is on the table’, ‘the house whose windows are dirty’.

77There are other pronouns (or pro-forms) that are concerned with gender, namely ‘some’, ‘any’ and ‘every’ compounds, as illustrated in the following pairs: ‘I can see something in the distance’ vs. ‘I can see someone in the distance’; ‘I can’t see anyone’ vs. ‘I can’t see anything; ‘I don’t know everything’ vs. ‘I don’t know everyone’.

78Some pronouns do not exhibit gender formally but are concerned with gender all the same in the sense that their use may be restricted to one type of referent (human or non-human). This holds for ‘one’, which normally denotes a human animate when it occurs in subject position (‘One is well aware that…’), unless there is some notion of contrast (‘I bought two dresses. One is blue, one is yellow’). Again the restriction only applies when ‘one’ has nominal status, not when it is a determiner (‘One person came’).

79Similar remarks could be made about ‘some’, ‘any’, ‘this’ and ‘that’, which in some uses are restricted to one type of referent (animate or inanimate). The grammar of pronouns is a vast subject that I can only touch upon (see Cotte’s paper in this volume for more details on the question of gender and pronouns in English). My only purpose here is to highlight the fact that pronouns are concerned with gender, and that this link with gender only applies to members of the nominal category. The question is now: why is it only the nominal category (understood in a broad sense, as including pronouns) that is concerned with gender? How do nouns categorize? And why would gender be apt to signify the type of categorization effected by nouns?

80Cotte (1999: 65-75) proposes that (grammatical) gender reflects the categorizing tendency of the mind. In his view gender is metalinguistic, i.e. expresses (in an abstract way) what nouns do. Several linguists have noted the separating role of nouns (Barsalou 2010, Cotte 1997, Langacker 1987, 1991 and 2008, Wierzbicka 1988), and what holds for nouns presumably also holds for pronouns. Langacker has stressed that ‘for nouns, the archetype functioning as category prototype is the conception of a physical object’ (2008 :103) (a physical object being typically bounded and separated from others) and that ‘a noun profiles (i.e. designates) a region in some domain’ (1991:15). I propose that the reason why gender aptly expresses the separation implied by nouns is that it is construed as an all-or-nothing opposition. We cannot be male and female at the same time, there are no entities which are both human and not human, we cannot be more or less male or female, nor can we be more or less animate or inanimate; gender is one of the best examples of non-gradability. The propensity of gender to express the cognitive process of categorization as effected by nouns can therefore be seen as coming from our direct experience of natural gender. As explained before (see section 2.4.) it is however the most typical situation (i.e. the all-or-nothing opposition, rather than the rare dubious cases that may arise) that is selected in the conceptualization of gender and that is used in a metalinguistic way.

81If gender expresses categorization, it is only logical that gender shifts (e.g. from non-human to feminine) may express processes of re-categorization. This is how I will interpret some cases where grammatical gender does not reflect natural gender.

4.3. A chronology of categorization

82In English it is normally the case that grammatical gender reflects natural gender. Gardelle (2006) notes that out of 2000 occurrences of she extracted from the BNC, not one applies to an inanimate referent. There are however some cases of mismatch between natural and linguistic gender, also noted in Gardelle (2006).

83I will focus on some atypical uses of ‘she’. We may sometimes (even though rarely), in some varieties of English, encounter the third person feminine pronoun in weather expressions (‘she’s raining’), and in some idiomatic expressions such as ‘she’s a toughie’ (i.e. ‘it’s a tough problem’), ‘She’ may also be used in the references to some physical objects such as boats, and for some institutions such as schools or universities. (This list is not meant to be exhaustive – these are the cases that I am going to address).

84In all cases I interpret ‘she’ as signalling a second stage in the process of categorization, for the following reasons.

85Gardelle, who mentions weather expressions occurring with ‘she’, also notices that this happens when the sentence describes extreme weather conditions: ‘she’s going to rain soup up there’, ‘she’s too cold to work on them plains today’, ‘Gosh she’s windy’ ‘and let ‘er rain’, ‘...let ‘er blow, let ‘er snow’ (2006: 212 ff). There seems to be no occurrence of ‘she’s sunny’ or ‘she’s pleasant’. I hypothesize that what triggers the shift from standard ‘it’ to ‘she’ is that the situation described is construed as exceptional, as more severe than what we would normally expect. Note in this respect the adverb ‘too’ (and the fact that the speaker has to change his/her plans), and the exclamation ‘gosh’. Also note in the first example the use of the noun ‘soup’ which provides a new (and unusual categorization) for an otherwise well-known phenomenon (rain).

86I also interpret as cases of re-categorization some occurrences of ‘she’ in idiomatic expressions such as ‘she’s a toughie’, ‘to whoop her up’ (i.e. ‘to brighten up a party’), ‘to touch her off’ (i.e. ‘set on fire’), ‘she’ll be apples’ / ‘she’ll be jake’ (i.e. ‘all is well, everything will be fine’ in Australian English), ‘to get her made’ (examples borrowed from Gardelle 2006). All these expressions may of course occur with ‘it’, but ‘she’ is also possible. What these expressions have in common is either the fact that a situation is seen as evolving (‘to whoop her up’, ‘get her made’, ‘touch her off’) or as re-assessed (there is an evolution in the construal of the referent). In ‘she’s a toughie’ for instance, a situation is first construed as being a problem, which is not what a situation is expected to be; this is the initial categorization of the referent. The situation is then said to be a tough problem: within the ‘problem’ category, it stands out because it is particularly difficult to solve. Note however that the expression is not ‘she’s a tough problem’ but ‘she’s a toughie’. The suffixed noun (‘toughie’) also indicates that the referent has been previously categorized, as it is only interpretable in a context where we already know that we are talking about a problem. About ‘she’ll be apples’ Gardelle explains that the expression probably comes from cockney rhyming slang. ‘Apple’ is a shortened version of ‘apple and spice’, which rhymes with (and therefore stands for) ‘nice’. The expression means ‘Everything will be nice / fine’ and is uttered in order to reassure someone. This means that a situation is first construed as difficult, and is then re-categorized (by someone else) as not so difficult. The re-categorization of the situation is echoed by the chronology of the denomination: ‘nice’ → ‘apples and spice’ → ‘apples’. ‘She’ reflects this chronology in the construal of the referent. Also note the use of the future; Gardelle (2006) notes that she has not found these expressions in the present or in the past (*‘She was apples’, *‘She is apples’). This also points to a process of re-categorization, which is anchored in time: the situation is a problem now but will not be a problem any more in the (near) future.

87Another motivation for the re-categorization may be the fact that several entities are grouped and then seen as one entity, which is construed as being more than the sum of its parts. This accounts for the ‘she’ that may stand for a university, a school, a church, a country (the U.S., the U.K.), nature, or even a language (as in ‘English as she is wrote’). In the examples given by Gardelle (2006:186-206), there are actually more possessive pronouns (‘her’) than third person personal pronouns (‘she’). This is consistent with the view defended here, i.e. that the referent is seen as a whole that transcends its parts; in order to be a possessor a referent must indeed be construed as a whole.

88A well-known case when ‘she’ stands for an inanimate object is that of boats. Here again I argue that this ‘she’ corresponds to a second stage in the categorization of the referent. Gardelle (2006: 191) notes indeed that only boats which have a name are referred to with ‘she’. (She also notes that, just as in the preceding cases, ‘her’ is actually more frequent than ‘she’.) It seems that when given a name boats acquire another status, which others do not have, this status being then reflected in the feminine pronoun. Note that this phenomenon is not restricted to boats: Gardelle (2006) gives the example of a ‘she’ standing for Elvis Presley’s guitar, which happened to have a name (‘Lucille’), and of the same pronoun used to designate the Rolls Royce’s emblem, named ‘Emily’.

89The feminine pronoun also seems to appear in contexts where something is separated from something else, as illustrated in the following examples borrowed from Gardelle (2006): (about a tree) ‘when I ripped her out of the ground’, (about a pot) ‘He jerked the pot from her bed’. I hypothesize that the fact that the tree, or the pot, is separated brings about a new categorization of the referent (a living tree becomes a dead tree).

90Note that in all the examples that I have discussed, when ‘it’ is not used for an inanimate referent, ‘she’ is used instead, not ‘he’. This reflects the general tendency noted by Gardelle (2006: 209): when ‘it’ is not used, the pronoun more likely to be used instead is ‘she’, rather than ‘he’. I interpret this as a means to emphasize the process of re-categorization. The feminine happens indeed to be the marked gender, in Indo-European languages broadly speaking but also in English: when gender is marked morphologically the feminine requires the addition of a suffix (‘-ess’ in ‘actress’ or ‘-in’ in ‘heroin’ for instance). The expression of the /+feminine/ semantic feature requires more signs (one additional suffix), an indication that the feminine is construed as an additional stage in the construction of the reference. The fact that the feminine is construed as a second stage in the specification of the referent is also visible in some pairs such as ‘horse’ vs. ‘mare’. The noun ‘horse’ holds for male and female horses, whereas ‘mare’ is only valid for a female animal. ‘Mare’ adds a semantic feature to ‘horse’.

91When a referent which is typically denoted by ‘it’ is denoted by another personal pronoun, the speaker has to make a choice between ‘he’ and ‘she’ (as there is no personal pronoun which only encodes the /+human/ feature). The feminine being construed as second compared to the masculine, it is only logical that ‘she’ is selected, as the feminine suggests in itself a second stage in the categorization process.

92Now the reverse situation, i.e. ‘it’ for a human referent, is also possible. Contrary to the preceding cases, it is neither rare nor unusual. It can be used to denote a figure (a silhouette) that one can see in the distance, or a stranger; it also occurs in (completely banal) sentences such as ‘It is John’. I have argued that ‘she’ signals a second categorization. I propose now that ‘it’ signals a first categorization. I therefore identify a system where pronouns (when they do not reflect natural gender) may signal different stages in the categorization of a referent, i.e. 1) ‘it’, 2) ‘he’, 3) ‘she’.

93Gardelle (2006: 216) notes that ‘it’ in reference to humans is frequent only when it stands for the noun ‘figure’. This lexical item is interesting as it suggests that one cannot really identify the referent (the referent may be too far, or the vision blurred). It therefore signals a first, incomplete, categorization, that may then be superseded by a better, more precise one.

94In sentences such as ‘it’s John’, or ‘it’s a boy’, I argue that ‘it’ also signals a first categorization. Cotte (1996: 208) comments on such utterances as ‘Somebody sat behind him. I saw that it was his brother’: ‘Un référent humain est normalement repris par he ou she. Mais il arrive que ce soit par it si on y réfère globalement, avant son identification (...). Chaque fois it dévalorise la première identification, souvent vague, pour valoriser la nouvelle identité focalisée en fin d’énoncé.’ What is most relevant for me here is that ‘it’ denotes a referent before it is identified. It seems that here again ‘it’ signals an initial categorization, which is expected to be followed by a more precise one.

95To come back to the sentences that are typically uttered when a baby is born, i.e. ‘It’s a boy’ or ‘It’s a girl’, we may wonder why the sentences *‘he’s a boy’ or *‘she’s a girl’ are not appropriate. The ‘it’ of these sentences may seem surprising, as this pronoun typically designates inanimates. One interpretation is that it neutralizes the male vs. female distinction. But what is the point of that, if what we want to convey is precisely the gender of the baby? Let us bear in mind that gender is conceptualized as taken for granted, as illustrated in the first two sections of this paper. Now what is it that is taken for granted in this situation? We do not know whether the baby is going to be a boy or a girl (so we cannot take that for granted), but there is one thing that we do know, which is so obvious that we do not even think about it, i.e. that the referent (of ‘it’) is a human being; in that sense the referent is pre-categorized. The pronoun that is used is ‘it’, the one that according to my hypothesis signals a first categorization. The first categorization in this case consists in the fact that we already know something about the referent, i.e. that it is a human.

96Now why is it the neutral pronoun, i.e. the one that typically stands for inanimates, that signals this first stage in categorization (in the chronology I have suggested)? I have already answered that question at the end of section 3.2.: ‘it’ typically stands for inanimates, and inanimates always precede human beings. We probably construe ourselves as second compared to inanimates, for the simple reason that the world existed before we were born.


97The gender of pronouns can of course be in keeping with the natural gender of the referent. In fact this is the most common situation in English. But when natural gender and grammatical gender differ, gender reflects stages of categorization.

98It is perhaps not so surprising that gender should express categorization, particularly the kind of categories denoted by nouns. In our experience we make a difference between humans (ourselves) and non-humans, and between males and females. Moreover, gender is an all-or-nothing opposition, particularly apt to express in a metalinguistic way the type of categorization expressed by nouns.

99We have a (by definition) complex concept of natural gender that includes several components (‘taken for granted’, ‘always already known’, ‘not talked about’, ‘all-or-nothing opposition’, ‘stable’, ‘inanimates come first’). However, when that concept is mapped onto language, only some facets (or even, sometimes, only one facet) of the concept may be used. The ‘already known’ component is used to signal anaphora, while it is the ‘stability’ one that is relevant when it comes to weather expressions. In sentences such as ‘it is John’ it is the fact that inanimates are construed as preceding humans that accounts for the use of ‘it’ to refer to a male human.

100In those uses it seems that grammatical gender has nothing to do with natural gender any more. What we map onto language is not gender as it occurs in the world but the concept of gender, or parts of that concept, the concept being eventually disconnected from its experiential basis.

Top of page



Barsalou, L., C. Wilson and W. Hasenkamp (2010). ‘On the vices of nominalization and the Virtues of Contextualizing’, in B. Mesquita, L. Feldman Barrett, & E. Smith (eds.), The Mind in Context. New York: Guilford Press, 335-360.

Benveniste, E. (1974). Problèmes de linguistique générale 1. Paris: Gallimard.

Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan (1999). Longman Grammar of Spoken and Written English. London: Longman.

Bolinger, D.L. (1977). Meaning and Form. English Language Series, 11. Longman: London.

Boone, A. and A. Joly (2004). Dictionnaire terminologique de la systématique du langage, 2nd edition (1st edition 1996). Paris: L’Harmattan.

Corbett, G. (1991). Gender. Cambridge : Cambridge University Press.

Corblin, F. (1995). Les formes de reprise dans le discours. Anaphores et chaînes de référence. Rennes: Presses Universitaires de Rennes

Cotte, P. (1996). L’explication grammaticale de textes anglais. Paris: PUF.

Cotte, P. (1997). Grammaire linguistique. Paris: Didier Erudition.

Cotte, P. (1999). ‘Le genre est une métalangue’, in Maret S. (ed), Féminin / Masculin, Littératures et cultures anglo-saxones. Actes du 38e Congrès de la SAES. Rennes : Presses Universitaires de Rennes, 65-75.

Dixon, R.M.W. (1982). Where have all the adjectives gone? And other Essays in Semantics and Syntax. Berlin and New York: Mouton.

Dixon, R.M.W. (1986). ‘Noun Classes and Noun Classification in Typological Perspective’, in Craig C. (ed.) Noun Classes and Categorization. Amsterdam: John Benjamins, 105-12.

Dixon, R.M.W. (2002). Australian languages: their nature and development. Cambridge: Cambridge University Press.

Gardelle, L. (2006). Le genre en anglais moderne (XVIe siècle à nos jours). Unpublished PhD Dissertation, University Paris - Sorbonne (Paris 4).

Hagège, C. (2001). Review of Unterbeck B. and M. Rissanen (eds.) (2000). Gender in Grammar and Cognition, in Bulletin de la Société de Linguistique de Paris, tome 96, fasicule 2, 78-79.

Halliday, M.A.K. and R. Hasan (1976). Cohesion in English. London: Longman, 1976.

Héritier, F. (1996). Masculin, Féminin. La pensée de la différence. Paris : O. Jacob.

Huddleston, R., G. Pullum et al. (2002). The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press.

Katz, J.J. and J.A. Fodor (1963). ‘The structure of a semantic theory’, Language 39: 170-210.

Katz, J.J. and P.M. Postal (1964). An integrated theory of Linguistic Descriptions. Cambridge Mass.: MIT Press.

Labrosse, P. (2003). ‘It dans un corpus de romans américains’, Etudes anglaises, vol. 4, tome 56, 482- 495.

Lakoff, G. (1987). Women, Fire and Dangerous Things. Chicago : University of Chicago Press.

Lambrecht, K. (1994). Information Structure and Sentence Form. Topic, Focus and the Mental Representation of Discourse Referents. Cambridge : Cambridge University Press.

Langacker, R. (1987). Foundations of Cognitive Grammar, Volume I. Theoretical Prerequesite. Stanford: Stanford University Press.

Langacker, R. (1991). Foundations of Cognitive Grammar, Volume II. Descriptive Application, Stanford: Stanford University Press.

Langacker, R. (2008). Cognitive Grammar: A Basic Introduction. Oxford: Oxford University Press.

Leech, G. (1981). Semantics. 2nd ed. Harmondsworth: Penguin. 1st ed. 1974.

Panaccio, C. (2011). Qu’est-ce qu’un concept? Paris : Vrin.

Rosch, E. (1978). ‘Principles of Categorization’, in Rosch E. and B.B. Lloyd (eds), Cognition and Categorization. Hillsdale NJ: Erlbaum, 27-48.

Rosch, E., C.B. Mervis, W. Gray, D. Johnson & P. Boyes-Braem (1976). ‘Basic Objects in Natural Categories’, Cognitive Psychology, 8, 382-439.

Rotgé, W. and Lapaire J.R. (2004). Réussir le commentaire grammatical de textes. Paris: Ellipses.

Saladin d’Anglure, B. (1977). ‘Iqallijuq ou les réminiscences d’une âme-nom inuit’, Etudes Inuit 1 (1), 33-63.

Saladin d’Anglure, B. (2010). ‘Un “troisième sexe” chez les Inuit (“transsexualité”, travestissement et chamanisme’ in Héritier F. (ed.), Hommes, femmes: la construction de la différence, Paris: Editions Le Pommier / Universciences Editions.

Senft, G. (2007). ‘Nominal Classification’ in Geeraerts D. and H. Cuyckens (eds), The Oxford Handbook of Cognitive Linguistics, Oxford: Oxford University Press, 676-696.

Shorter Oxford English Dictionary on Historical Principles. Fifth edition (2002). Oxford: Oxford University Press.

Stalnaker, R. (1973). ‘Presupositions’, Journal of Philosophical Logic 2, 447-457.

Taylor, J.R. (2003). Linguistic Categorization : Prototypes in Linguistic Theory. Oxford : Oxford Clarendon Press.

Van Hoek, K. (2007). ‘Pronominal Anaphora’ in Geeraerts D. and H. Cuyckens (eds), the Oxford Handbook of Cognitive Linguistics, Oxford: Oxford University Press.

Wierzbicka, A. (1988). ‘What’s in a noun? or: How do nouns differ in meaning from adjectives?)’ in Wierzbicka A., The Semantics of Grammar. Amsterdam: John Benjamins Publishing Company, 463-497.


Allen, W. (1983). Four Films of Woody Allen. London, Boston: Faber and Faber.

Coe, J. (2007). The Rain before it Falls. London: Penguin Books.

Sillitoe, A. (1959). The Loneliness of the Long-Distance Runner. London: Flamingo Modern Classic, Harper Collins Publishers.

Top of page


1 I underline the elements that are relevant for my analysis.

2 This remark is based on personal observations. It is very difficult to test empirically in large electronic corpora such as the CHILDES corpus, for two reasons: 1) there aren’t any search tools allowing for an onomasiological approach, i.e. one can’t look for contexts where gender is predicated; 2) the nature of the linguistic interaction is fairly restrictive: all conversations are held between an adult and a child, which probably yields different types of conversations than the ones that spontaneously occur between children.

3 My translation from French of personally collected data (sentences uttered by four-year-olds).

4 Mummy and I are both... hmm... women... well... no… I’m not a woman. I’m a girl and Mummy’s a woman (translation mine).

Top of page


Bibliographical reference

Elise Mignot, « The Conceptualization of Natural Gender in English », Anglophonia/Sigma, 16 (32) | 2012, 39-61.

Electronic reference

Elise Mignot, « The Conceptualization of Natural Gender in English », Anglophonia/Sigma [Online], 16 (32) | 2012, Online since 24 December 2013, connection on 19 December 2018. URL : ; DOI : 10.4000/anglophonia.140

Top of page

About the author

Elise Mignot

Université Paris-Sorbonne, CELTA, E.A. 3553.

By this author

Top of page


Licence Creative Commons
Anglophonia – French Journal of English Linguistics est mis à disposition selon les termes de la licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International.

Top of page
  • Logo Presses universitaires du Midi
  • OpenEdition Journals