Pronouns and gender

In modern English, the third-person-singular pronouns are the only ones differentiated by gender. Was this always the case? Historically, how did this come about?

Yes.

Old English 3rd-person pronouns were gendered the same way as in modern English (even if some of the words were different)

Singular:
feminine heo
masculine he
neuter hit

Plural
all genders: hie

She was not originally a personal pronoun but derived from the OE feminine demonstrative pronoun seo meaning ‘that one’ or ‘the’ (f.)
They originated as a loanword from Old Norse þeir which was a masculine plural pronoun, both personal and demonstrative.

I’m just sneaking in to say that in Arabic, the second-person pronouns are differentiated by gender. Both versions are written the same, أنت, but the male one is pronounced anta while the female one is anti (it’s possible to vocalize texts, in which case the two variants are written differently).

In the plural, the two pronouns differ also in writing - it’s أنتم (antum) for male and أنتن (antunna) for female.

But I’ve never seen this differentiation in an Indoeuropean language.

Wouldn’t it depend on how far back one went in time? I think the OP is asking when did this change come about from the original gendered form in whatever ancestral language last had gendered forms.

Not in this case. In Proto-Indo-European, first- and second-person pronouns don’t seem to have ever been marked for gender. (Number and case, yes; gender, no.) The third-person pronouns are originally demonstratives, which is why they are treated differently.

ETA: This distinction seems to be older than the development of the three-class grammatical gender system (masculine, feminine, and neuter) in PIE, so I think in this case we really can say it has always been thus.

It’s doubtful that the earliest form of Proto-Indo-European had the 3 genders. Because the earliest language to separate from PIE was Hittite, which had no feminine gender. Instead, Hittite nouns were divided into animate and inanimate noun classes. One theory is that after Hittite separated out, the remainder of PIE then developed the familiar 3 genders.

This page gives a version of perhaps the most prevalent theory of IE gender origin–

Quite fascinating. There may be traces of the hypothetical animate/inanimate system to be found in IE languages.

There are often two PIE roots reconstructed with the same meaning-- for example *igni- and *pur- both meaning ‘fire’.
In the daughter IE languages that kept gender, the reflexes of *igni- are always masculine: Latin ignis, Lithuanian ugnis, Russian ogon’, Sanskrit agni. (Although I don’t know what’s going on with that anomalous unstable initial vowel!)
Meanwhile the reflexes of *pur- are neuter: German Feuer, Greek pyr, Old English fyr.
According to this hypothesis, the animate nouns became the masculine gender, while inanimate became neuter. As if *igni- referred to the fire as a living being (leaping flames?) while *pur- was for fire seen as an inanimate thing (embers?). After all, it was Agni who was the Vedic fire god.

More about pronouns: The PIE demonstrative pronouns were *so, **seh[sub]2[/sub]*, and *tod, becoming Old English se, seo, and þæt. The latter two of these are the ancestral forms of “she” and “that.” Originally, they all meant ‘that’ but now the neuter/inanimate form has become the only one used. “She” has shifted to take the place of the feminine 3rd-person pronoun.

The way demonstratives replace 3rd-person pronouns can be seen in many languages, e.g. the original Mongolian 3rd-person pronoun i was lost very early in the history of the language and replaced by ter meaning ‘that’. Likewise in Hindi, the only way to express the 3rd-person pronouns is to use the words for ‘this’ and ‘that’. Which have no gender. Hindi is the opposite of English that way: nouns and adjectives have gender, but pronouns don’t.