Just here you said that a male might have just masculine traits. Could a female have just masculine traits? Or does the definition of 'female' preclude that? — bert1
I think the problem here is that it's not clear what it means to have "traits". Phenomenological constructivism says people construct a world view, and within that you make sense of the world in a way that fits what you do. So what you have to ask yourself is how how it would be possible for someone to have only "masculine" traits to be constructed as female within this world view.
Stuff you take for granted is only going to enter awareness when it becomes problematic. My initial hunch is that if you see someone who "only has masculine traits", you're likely going to just assume male, as some sort of "background radiation" of meaning.
But even more importantly: by the time we notica "feminine" trait about a "man", we've already categorised the "man" as "man", and that's why a "feminine" trait is remarkable enough to enter awareness and even focus as feminine in the first place.
However, for a sociologist, the question is primarily empirical. If you do think of someone as female, but when you try to assign traits you can't think of a single stereotypically masculine one, why is that? That would be a very interesting situation that could lead to refine the particular theory.
Note that we think of, say, a penis as a male attribute, not a masculine one. That's a social coding of some sort, part of what we think of fundamental. The thing is, though, that usually genitals are not on display, and not all people wear clothes that make the genral contours plain in view, yet, for the most part we slot people into male/female with no hesitation. We're probably wrong about that categorisation now and than without ever noticing, because attention is fleeting when it comes to passes-by. And we only notice that we're constantly assigning gender, because the process sometimes fails, and we take a second look to figure that stuff out. So the stuff that comes up in discourse about whether or not the trans condition is real, like genitals or DNA, are probably not the traits we primarily use in day-to-day life to make those judgments. They're tie breakers that work as long as worldviews are compatible.
So let's start with your (2): how people feel inside. Children acquire their worldview while living with their parents and peers (and their wider social circle beyond that). At some point they probably acquire a sense of what it means to be boy or a girl. But they've been gendered in other worldviews before that: and if you've just gone along with how people who gendered you in <i>their</i> world view treat you then your behaviour is going to be compatible with that worldview at the time you acquire the distinction in question, and the distinction is just one among many largely unproblematic facts about the world you take in. That does not mean that you can't sometimes "buck the trend". You can be a boy and play with dolls, for example. Depending on your parents views on propriety, you're going to run into different amounts of "trouble", the lightest probably being a short moment of surprise, or maybe even none at all. The degree to which a boying playing with a doll is noteworthy, is the degree to which the act is stereotpically feminine. You don't get an affirmative reaction to a girl playing with a doll in the same way, unless you've been "worried" that she's not "sufficiently" feminine. But all of this occurs on a baseline of maleness and femaleness. And that's your (1).
Appearance matters quite a lot when it comes to gender assignment, and if we're not sure we have couple of more privacy-intrusive methods to check: genitalia, DNA, etc. Biology. But the way we look at biological sex is heavily influenced by our interest in the topic. The categories we use to describe sex are inevitably gendered.
Trans people are, compared to cis people, very rare. They know they're transe because of how they feel inside, but that's hard to communicate, because other, much more common world views don't include that sort of discrepancies. So to figure out what to look at when it comes to biology you'd need to listen to them, but to listen to them you'd need to take them seriously, and accept that your failiur to understand is your failure to understand, and not, say, a delusion of the person who feels something - to you - incomprehensible. Someone who responds to "I'm a trans-man, you're a cis-man," with "that's stupid; you're obviously a woman," isn't likely to be in favour of funding research as to the biology of trans people.
It's not that what what we know about sex is wrong, it's because sex is gendered through a cis bias, that we the categories we have to describe sex are insufficient for the needs of trans people. I've looked into then recent research at some point, and thought that was interesting, but I'm not enough of a bilogist to understand that sort of stuff easily, and I haven't retained much. But there's definitely a gendered compenent to how we research sex and what we look at. So when you conclude here the following:
It is senses 1 and 2 that determine a person's gender, and sense 3 only adds masculinity and femininity to that. So what I'm questioning is that sense 3 is not really about the male/female opposition, and wholly about the masculine/feminine opposition. — bert1
I'd say that we (cis-people) are used to use (1) to legitimise our gender, but it's really (1), too. But because we're the majority it stands largely unchallenged and doesn't often our awareness. A trans person (and a genderfluid and agender person) would be more aware of (3), simply because they keep clashing against the mainstream. The main struggle is not to be agreed with; it's to be understood in the first place, or even to get people to realise that they're misunderstood. And it's difficult to talk about because the gendering of sex also heavily influences our vocabulary. That's how we get the new prefix "cis-". But it's difficult to promote the term when cis-people generally don't have the experience that pushes the entire problem-area into awareness. There's a whole baseline of how some people relate to their body that we can't intuit. The same is true, presumably, for trans-people: what is it like to blind to that area? The difference is that nearly everyone they meet will fall into that category, from childhood on. I've heard time and again what a relief it is to find other people with a similar experience.
Gender then is the entire constellation. What and how many constellations do we find meaningful? What do we attribute to biology, etc. A social construct tends to only enter awareness if it's problematic, and the mainstream gender conception becomes problematic when we ponder trans people, intersex people, other constellation in other species (write a SF story about sentient slime molds, for example), other orderings of the same biological matter of facts in other cultures or sub-cultures etc. A social constructivist would abstract (3) from a set of compatible world-views, I think. (It's even more complex, because gender is only part of any given worldview, and worldviews might otherwise be largely compatible.)
I'm probably not explaining this very well, since I'm... unsure myself. It's been more than 20 years since I read any of the literature. I left university in my late 20ies and I'm now nearly 50. And it's really hard to understand in the place, because sort of have to imagine a world-as-is beneath a world-as-experienced, while also maintaining that you can't really do that.
Think about animated films for example: Robots, Cars, Brave Little Toasters... they're all gendered, without, logically, having a sex. We create the illusion of "sex" with very few signals, without actually assuming the underlying biology (since none of those have an underlying biology). How does that work? Gender is a sort of narrative we use to explain sex: without gender, sex esists, but is meaningless. Does that make more sense?