By the way, regarding the topic of the thread...
What I would like to know is - and please excuse my ignorance of Latin - is there any justification for this rule in Latin grammar? Are present-day purists who oppose the reform of the French agreement rule by appealing to Latin misguided?
Human languages sometimes do develop rules that are based on whether something is found before or after a certain word. For example, in Spanish and much of Romance, there is a pattern that a definite direct object noun phrase that is topicalized to the front (the beginning of the sentence) is often mentioned again with a direct object pronoun.
Descubrí el escondite. 'I found the hideout.'
El escondite
lo descubrí. 'I found the hideout.' (with
el escondite 'the hideout' topicalized to the beginning)
Similarly, in Spanish there's a rule, by all means natural and followed in spoken language, that negative pronouns and adverbs found
after a main verb trigger negative agreement in the verb (i.e. the verb must be negated, with
no), but if they're found
before the verb, then they don't.
Nunca corro. 'I never run. (~ I'm never in a haste.)'
No corro nunca. 'I never run.' (with post-verbal
nunca triggering
no next to the verb
corro)
En nada pienso. 'I'm thinking of nothing.'
No pienso en nada. 'I'm thinking of nothing.' (with post-verbal
nada 'nothing' triggering
no next to the verb
pienso)
In English, even if the construction itself is a bit old-fashioned now, clause-initial negative words trigger (or used to trigger) the usual switch between do/have/be and the subject, even though no question is implied.
I didn't do it, and I didn't think it was right either.
~ I did not do it,
nor did I think it was right.
I have never seen anger like that.
~
Never have I seen such anger.
I don't find it weird that French developed a rule on similar lines for participle agreement in compound tenses. It is unfortunate that this rule still exists now, though, since most French speakers pronounce all forms the same (aimé, aimée, aimés and aimées sound the same for most speakers, especially outside Belgium), and even for those participles that can sound differently (pris vs. prise/prises), many if not most people don't do it naturally anymore anyway.
However, the agreement rule is also found in Old Spanish. Compare the following quotes, taken from the 12th-century
Poem of the Cid. Here's a couple lines where the direct object is found after the verb, so there's no agreement in the participle:
Vençid
o a esta batall
a (line 1008, no agreement; if it did: *
vençida (h)a esta batalla)
'He has won this battle.'
Dexad
o ha heredades & casas & palaçios (line 115, no agreement; if it did: *
dexadas ha heredades & casas...)
'He has left [his] inheritance, houses and palaces.'
Meanwhile, if the direct object is found before the verb (below:
ha dexado, han tornado), then the participle agrees in gender and number (
dexada, tornados):
Asos castiellos al
os mor
os dentro los an tornad
os (line 801, cf. modern Spanish
a los moros han regresado adentro de sus castillos)
'They have turned the Muslims back into their castles.'
vna tiend
a a dexad
a (line 582, cf. modern Spanish
una tienda ha dejado)
'He has left one tent behind.'
And again, as Pacifica just said, in Latin, past participles
always agree in gender and number, so French and Old Spanish (and to an extent Italian) have simply constricted that agreement to some grammatical contexts. Note that although Portuguese has lost all participle agreement in its
ter + participle construction (
os castelos têm abandonado), Spanish retains it in all contexts in
tener + participle (
los castillos tienen abandonados). (Although in Spanish, this construction is less common and less basic/essential than the equivalent in Portuguese.)