English's lack of inflectional morphology has consequences. One of them is the attendant need for a fairly rigid syntax, the structural position of morphemes within a sentence, to convey the necessary information needed to decode the sentence. If you're writing a poem in Latin, you have remarkable freedom to move words around within a sentence, as (usually) the inflections carry the information necessary to determine the relationships within the words. Case markings, for example, denote subject-object relationships. But in English, we only have case markings for pronouns (which here I will use to refer to personal pronouns)-- he/him, she/her, I/me. (In fact English had a broader case structure, but it was lost in the transition from Old English to Middle English; a vestigial example is found in the slowly dying "whom.") We also only have gender in singular pronouns and not in our verbs or nouns that aren't semantically assigned to one gender or sex. (By semantic gender I mean referring to actual gender or sex in the real world, and not purely linguistic gender, as in the assignation of gender to inanimate objects in languages like Spanish.)
Plural pronouns lack gender inflection, given that referring to any group might necessarily include people of both sexes. (Although note that there's no reason we couldn't have a plural pronoun that referred only to a specific sex or gender.) You/your, we/our, they/their-- each is plural and gender indeterminate. Gendered possessives are found only in the singular-- he/his, she/hers. We can say, therefore, say "every girl threw her ball" or "every boy baked his cake" without trouble. But what about possessives that consider both genders? What do we make of a sentence like "Every child lost their marbles?" It's a question I considered for my semester project in a class in generative grammars and minimalist syntax last semester. I'm not a linguist myself, though I am in a related field, so please take this post with the requisite grains of salt. Here's what I found.
We have this problem in English: we're lacking a particular pronoun, the third person gender-neutral singular. The conventional way around this is to use their: "every student picked up their paper." But this usage drives prescriptivist grammarians crazy, as "every" is singular, which we can tell from how "student" inflects as singular. (It's "every student," after all, not "every students.") The typical advice is to instead us "his or her" in place of their. That's a technically satisfying answer, but as anyone who actually uses English knows, it's imperfect: it sounds clunky, likely due to its phonological distance from the other possessive pronouns like I, me, she, he, you-- each of which has only one syllable. That's not an irrelevant concern. Phonological symmetry is actually an important consideration when it comes to certain categories of words.
Urs Weidman, in a chapter from a 1984 edited collection, called this missing pronoun "Pronoun X." He defined its features as follows:
◦Syntactic Class: Determiner/Pronoun
◦Prosodic quality: Unstressed
◦Semantic Class: Human
The semantic class of "human" listed there is important, as it forbids the use of "its." "Its" carries no gender distinction but is not eligible to refer to human subjects. As Weidman points out, there have been historical efforts to generate the missing pronoun in order to solve this problem-- "ho," "thon," "he'er," etc. Unfortunately, the determiners (a syntactic category of which this type of pronoun is a part) are a functional category-- that is, they are used primarily to structure language and not to deliver semantic or propositional content, which is the role of the lexical categories. You can invent new verbs and nouns-- we do it essentially every day-- but not, say, new articles. So these replacements have never caught on, and lacking the necessary term, we've regularly used "their."
Pronouns, Anaphors, and Binding Theory
To grasp what's happening with singular "their," it's necessary to talk about pronouns, anaphors, and antecedents, and a relationship dictated within them called binding theory. Binding theory describes the conditions under which certain pronominals and anaphors (such as himself/herself, etc.) can occur in English. When two terms are semantically linked to indicate the same real-world reference, such as in "Freddie deBoer" and "me," we say they are coindexed. Terms that are coindexed can be broad/grouped/undefined, but they still refer to some consistent referent.
How does binding theory operate? Binding theory dictates situations in which two terms are coindexed and one c-commands the other. C-command is a structural relationship in syntax that helps determine a broad range of operations; understanding c-command requires an understanding of syntax tree structure that isn't really important for what we're doing here. If you'll forgive a bit of squishiness, it'll be easier to talk about subject-object relationships. So one rule of English binding theory is that it permits reflexive anaphors to be taken as objects by their own coindexes-- "John loved himself"-- but not personal pronouns-- "*John loved him," where "him" refers to John. Similarly, pronominals can't take their own coindexes as objects-- "*He loved John," where "he" refers to John. So far, so good.
Antecedents, variables and the problem with "gender neutral"
I've used the terms "gender neutral" or "gender indeterminate" to refer to the use of plural possessives like "their." But a little investigation reveals that we don't actually get the "their" problem with every gender neutral antecedent. For example, a term like "student" doesn't provoke that construction by itself; we don't see people saying "The student picked up their paper" where "their" is coindexed with "student." What provokes the construction, rather, is the presence of a quantifier, such as "each," "every," and "some." Those particular terms are unusual in that they are syntactically singular but refer to a range of possibilities-- "each student," after all, refers to students as individuals but applies to a group of individuals.Those are the conditions under which we are likely to get the problematic "their."
What we're talking about here, technically, are bound variables. Pronouns are variable because they can refer to different real-world referents (that is, they can be indexed with different terms, depending on context). Typically, pronouns don't require a specific syntactic position. You can say "when did he get here?" without ever having said who he is. This makes them free variables. Anaphora, however, require a certain locality to their antecedents. You can't just sprinkle "himself"s just anywhere. They therefore are bound. The kind of antecedents that I've been referring to, the quantifier-gender neutral noun combinations like "each student," are bound because they have a necessary structural position (c-command) and variable because they refer to no particular individual. They are syntactically limited but semantically fungible, an unusual case. (A deep consideration of bound variables by Barbara Partee can be found here in PDF form.)
The Syntactic Conclusion
My final conclusion, then, was as follows: “Their” with a singular coindex is a bound variable that must be c-commanded by a Spec where the DET is a quantifier that merges with a singular NP that does not designate a particular gender. That is, we see the use of "their" as a singular possessive pronoun only in cases where the object using "their" follows a subject which contains a quantifier that inflects as singular but refers to any individual of a possible range of choices. I've provided a syntax tree of such a sentence below.
The Grammar Wars
All of that is long-winded and technical. The practical question persists, however: what to do about "Everyone needs to keep their hands inside the moving vehicle at all times"? The syntactitians have it easy; their definition of grammaticality is simply "that which is recognized by a native speaker as a possible utterance," and considering that this type of construction is voiced all the time, it qualifies. But prescriptivists grammars hate it, because of the pairing of a plural pronoun with a singular antecedent. On the other extreme, the living language crowd says that anything goes.
I am not, myself, either a rigid prescriptivist nor a pure "anything goes" type. The plain fact is that language lives and evolves, and trying to hold onto one version of the language is like trying to grab a puff of smoke. Yet I also think that simply saying "hey, language evolves, so get over it" is too easy, and threatens our ability to use language in a specific and nuanced way. Too often, in my estimation, the anything goes types actually become their own kind of prescriptivists, with whatever current construction they like becoming the right way. After all, if anything goes, the traditionalists are perfectly free to prefer the stricter interpretation. And if we want to be adaptive and cosmopolitan, we shouldn't borrow prescriptivist readings when it suits us. When dealing with scolds, it's nice to be able to point out that "they" was used as a singular pronoun for centuries before anybody said that you couldn't. But we shouldn't be tempted to take that as dispositive when we are trying to avoid exactly that kind of rigidity.
For myself, I try to keep my scoldings about grammar and vocabulary limited to the cases where there seems to be a genuine threat to meaning and specificity. For example, I would never let a "your/you're" problem go in a paper I was grading; there are a myriad situations in which such substitution can obscure meaning. I'm also touchy about the meanings of many words, as lexical drift tends to move away from specificity, and specificity is the key to writing well. I don't want "literally" to become some meaningless intensifier because when it does, I no longer have a word that means "in the narrow denotative sense." When "ironic" comes to mean simply weird or funny, a particular and important idea is lost into a vague mush. Some things I'm willing to let go of, and some I'm willing to fight for.
Using "their" for singular antecedents is one that I think people need to just give up on. As I've argued, it only occurs in a very limited set of circumstances, and those circumstances uery unlikely to produce confusion about what is meant. We all know what is intended in such a statement, to the point that most of us don't even notice it in spoken conversation. And as we lack a satisfying alternative, the usage is likely to persist. That's not to say that you shouldn't understand what the "rule" is, if only to be able to satisfy those gatekeepers that police it. (Don't use it in your resume, don't use it in your grade school application.) But this is an example of a gate that's not worth defending anymore. I no longer correct it on my own students' papers. Those who want to persist with "his or her" can go right ahead, but perhaps it's better to leave others alone. Everybody can make their own choice.
Beletti, Andriana and Luigi Rizzi. “Psych verbs and Θ-theory.” Natural Language and Linguistic Theory 6 (1988): 291-352. Print.
Carnie, Andrew. Syntax: A Generative Introduction, 2nd edition. Malden: Blackwell, 2007. Print.
Chomsky, Noam. “On binding.” Linguistic Inquiry 11.1 (1980): 239-245. Print.
---. Lectures on Government and Binding. Dordrecht: Foris, 1980. Print.
Evans, Gareth. “Pronouns, Quantifiers, and Relative Clauses.” Canadian Journal of Philosophy 7.3 (1977): 467-536. Print.
Partee, Barbara. Compositionality in Formal Semantics - Selected Papers by Barbara H. Partee. ed. Barbara H. Partee. Malden: Blackwell,1978. Print.
Radford, Andrew. Syntactic theory and the structure of English: A minimalist approach. Cambridge: Cambridge UP, 1997. Print.
Weidmann, Urs. “Anaphoric They for Singular Expressions.” Modes of Interpretation. Zurich: Gunter Nagg,1984. 59-70. Print.