What do we mean when we say that a certain human (or even non-human animal) group X is marginalised (in a context C)?
The expression is ambiguous.
It could be a descriptive claim, meaning that
A. A (generic) X is at a disadvantage with respect to a (generic) non-X (most commonly a hegemonic group Y) in a common context C.
For example, to say that women are marginalised in Mexico might mean that women make less money that men, are less represented in positions of power, receive less education and health services, etc.
But it could also be an explanatory claim, meaning that
B. A (generic) X is at a disadvantage with respect to a (generic) non-X (most commonly a hegemonic group Y) in a common context C because she is X (and not Y).
In other words:
B. A (generic) X is A-marginalised (with respect to Y and C) because she is an X.
[From now on I am going to obviate saying that these claims must be read as generic and with respect to a hegemonic group in a common context. Also, in theses B onwards, marginalisation will mean A-marginalisation]
Weak-B. At least some of the causes of why the Xs are marginalised are that the Xs have certain properties that the Ys do not or they have them more frequently.
Prima facie, A is very good, yet defeasible, evidence of B. It would be quite a coincidence if the Xs were marginalised for causes that are more or less evenly distributed on the Xs and the Ys. However, some people take B to be a stronger claim:
Strong-B. At least some of the causes of why the Xs are marginalised are actions intended to marginalise the Xs.
Even though A is good evidence of weak-B, most current theories of marginalisation recognize that neither A nor weak-B offer but very weak evidence of strong-B.
Between weak-B and strong-B there is another thesis of marginalisation that has been important in recent debates on marginalisation:
C. A (generic) X that passes as Y is less marginalised than a (generic) X.
The idea is that passing for a member of a hegemonic group ameniorates some of the marginalisisng effects of belonging to a marginalised group. For example, gays that pass as heterosexuals are less discriminated that gays that do not. It is not obvious how to best characterise passing, for it depends on what it means to be an X or a Y. For example, it can mean to show the appearence of being a Y, or something stronger like perfoming the values of Y. Borrowing an example from Knobe and Prasada, one can say that Part of why Hillary Clinton has been able to rise as one of the main presidential candidates has been because she embodies masculine values and traits.
Now, I say that C stands between weak-B and strong-B because strong-B gives us good but defeasible evidence of C and of course, C, like strong-B, entails weak-B.
C is also important because it is usually linked to a different set of claims regarding the marginalisation of X. So far, A, B and C are all individualistic accounts of marginalisation. In other words, in them, for Xness to be marginalised is nothing above the Xs being marginalised. However, for Xness to be marginalised can be conceived in a non-individualistic way. In other words, so far X-ness has been conceived as a property (or set of properties) that individuals have or have not. But Xness can also be conceived as something else that emerges from the joint action and customs of the Xs. This anthropological sense of Xness is most obviously and readily applicable to ethnic groups, where Xness refers not to any property that the X have but to the links that bind them together as Xs, i.e., their cultural products, customs, etc.
For example, in a recent interview American celebrity chef Anthony Bourdain expressed concern that when people think of haute cousine they do not consider Mexican food, even though there is very good Mexican haute cousine. Indeeed, in Mexico, and outside Mexico as well, Mexican food is strongly associated with “antojitos”, that is, comfort food and snacks, and not with haute cousine. Now, since comfort food and snacks are commonly valued less that haute cousine, this marginalises not (or not only) Mexicans as individuals, but Mexicanity itself.
Thus, we have:
D. Xness is devaluated (over Yness in a common context C).
As I mentioned, D-marginalisation is clear in cases of ethnic marginalisation, for example, of Kurds in Turkey, where there are explict measures to diminish the use of the Kurdish language and other similar cultural manifestations. However, it does not apply as easily to other marginalised groups, for example, women, because women lives are more tightly interwoven with the lives of non-women – not just because they live together, but because they commonky build string bonds of caring and other affects with them – that the lives of members of one ethnic group are related to the lives of members of a different ethnic group. To paraphrase Paloma Hernández, it is more common for a woman to have a non-woman child or life partner that an African-american having a non-African-american child or life partner.
D entails C, but D cannot be reduced to C. C is more of how D manifests at the individual level.
Furthermore, even though D generally causes A, it remains an open question whether, when they happen together, A or B is a more fundamental form of marginalisation.