Timeline of future/conditional in Latin and Romance languages

Question

I'm not a linguist - just a linguistics enthusiast - so apologies in advance if this is a stupid question.

I am fascinated by the concept of grammaticalization, and I had heard that the future and conditional in many Romance languages give a relatively clear example. As far as I understand, the modern inflected forms began as the infinitive followed by an appropriately conjugated auxiliary verb, and over time the auxiliary fused with the infinitive and possibly underwent some sort of sound change. The stages would be something like (using Spanish, since I don't know Latin)

1. Figurate	2. Analytic conditional	3. Fused/synthetic form	4. Sound change/erosion
saltar había	saltar había	saltarhabía	saltaría
Figurative, but universally understood	No longer seen as figurative, but still analyzed as infinitive + conjugated auxiliary	No longer analyzed as infinitive + conjugated auxiliary	Loss of middle consonant of haber

(I'm not sure about the ordering of 3 and 4.)

The question: Something like the process above happened in the history of each modern Romance language (that has this construction). Where along this process should one place the split of Latin into distinct Romance languages? More than the answer itself (which is really a matter of history), I'd like to know, in as much detail as possible, what one can deduce based on knowledge of modern Romance languages only.

---

My thoughts:

Before 1. It seems like the problem with putting the split before the process even begins is that it would require at least 5 Romance languages (Catalan, Spanish, French, Italian, Portuguese, etc.) to have separately developed the same peculiar construction for expressing future action: infinitive + "to have". On top of this, haver/haber/haver aren't even used in Catalan/Spanish/Portuguese to express possession the way avoir/avere are in French/Italian (though maybe they once were?), so the semantics at the time of creation of this formula would not have been the same.

After 4. I can't tell for sure, but something I read seems to suggest that the split actually happened after Latin had developed something like (a Latin version of) "saltaría", and then it differentiated, giving the various Romance versions, but then I don't understand how one gets the following for, eg., the future:

Spanish		French		Italian
saltar	haber	sauter	avoir	saltare	avere
saltaré	he	sauterai	ai	salterò	ho
saltarás	has	sauteras	as	salterai	hai
saltará	ha	sautera	a	salterà	ha

If by the time the Latin split into Spanish/French/Italian, speakers no longer understood the future endings as present indicative conjugations of haber, how did the endings manage to stay exactly in sync with those conjugations?

It seems like one needs to posit a sort of hybrid mental state for speakers, where the future/conditional had become synthetic enough that erosion of the 'b' was allowed, but were still analytic enough that the endings were recognized as present/imperfect conjugations and evolved as such in each language - actually, not just any present/imperfect conjugations, but specifically those of haber, which is irregular in the present in Catalan/Spanish/French/Portuguese.

I would really appreciate if someone could help clarify these issues!

Answer 1

The Conditional and Neo-Future are present in the vast majority of the Romance languages.

To my knowledge the only Romance languages which lack them are Sardinian and the various Balkan Romance varieties (e.g. Romanian, Aromanian, Istro-Romanian, and Megleno-Romanian). These varieties broke from the Common Romance dialect continuum at a very early stage as evidenced by their lack in participation in many other later Common Romance changes, such as the evolution from a five-vowel-quality two-length system to a seven-vowel-quality system (in stressed syllables), and the lack of Palatalisation of velars before front vowels in Sardinian.

We should also note that European Portuguese (and also Galician) exhibit the phenomenon of mesoclisis, where the clitic marking a pronominal object is placed between the infinitive stem and the ending (derived from the auxiliary) e.g. European Portuguese dar-mo-ás (regularly contracted from what would otherwise be *dar-me-lo-ás) vs Brazilian Portuguese & Spanish me lo darás.

I've often seen it said this occurs only in formal registers, but according to Janus Bahs Jacquet's comment to this answer here it's more of a dialectal difference, being present in European Portuguese (and Galician) even in informal speech, but absent in Brazilian Portuguese.

With those facts in mind, it seems likely that the main parts of the process took place in the Common Core Romance Period (after the divergence of Sardinian and Balkan Romance, but before the Core Romance dialect continuum broke apart), with the fusion of the ending (from the auxiliary) to the infinitive stem being a slightly later development that spread to all of Core Romance other than Galician and (European) Portuguese.

So the process is something like what follows:

In Latin proper, the future tense of saltō "I jump" is saltābō, saltābis, saltābit, saltābimus, saltābitis, saltābunt. There is no conditional.

In Common Romance, the Latin future tense is lost. Note that the latin future tense of the first and second conjugation is itself a compound tense formed from the present stem followed by a clitic derived from a reduced form of the present tense of PIE **bʰuH*- (verbs of all conjugations use the imperfect of this clitic to form their imperfect, something that is true of all Italic), whence also Latin fīō "to become" (the third and fourth conjugations instead take their future forms from the PIE subjunctive).

After Sardinian and Balkan Romance split off from Core Romance, a new analytic future develops consisting of the infinitive followed by a cliticised form of habeō "I hold", giving the future tense saltarɛ-aβɛʊ, saltarɛ-aβɪs, saltarɛ-aβɛt, saltarɛ-aβɪmʊs, saltarɛ-aβɪtɪs, saltarɛ-aβɛnt (noting regular sound changes to Proto-Core-Romance, e.g. lenition of intervocalic b & h, as well as vowel shifts).

By putting this clitic into the imperfect, a conditional was also formed saltarɛ-aβɪβa, saltarɛ-aβɪβas, saltarɛ-aβɪβat, saltarɛ-aβɪβamʊs, saltarɛ-aβɪβatɪs, saltarɛ-aβɪβant.

The auxiliary then reduces in slightly different ways in Italo-Romance (which preserves some β's, as seen in the 3rd person conditional -èbbe and -èbbero that attach to the infinitive, or the gemination in the 1st person plural conditional -émmo also attached to the infinitive), and the rest of Romance, which lose all the β's, and vowels that end up in hiatus also contract in various ways in distinct varieties. This is also likely the point at which the stress moves from the infinitive onto the auxiliary.

At this point we have a Western Romance future something like saltar-aɛʊ, saltar-aɪs, saltar-aɛt, saltar-aɪmʊs, saltar-aɪtɪs, saltar-aɛnt, and a conditional something like saltar-aɪa, saltar-aɪas, saltar-aɪat, saltar-aɪamʊs, saltar-aɪatɪs, saltar-aɪant, with the auxiliary still a separable clitic.

Following the contraction of the many hiatuses, this gets us to the situation present in Galician and (European) Portuguese.

E.g. in Portuguese we have the future saltar-ei, saltar-ás, saltar-á, saltar-emos, salta-eis, saltar-ão, and the conditional saltar-ia, saltar-ias, saltar-ia, saltar-íamos, saltar-íeis, saltar-iam. Note that the hyphen here is not present in the orthography, but serves to show the position pronominal object clitics can be placed, showing that the auxiliary is still a clitic in these varieties, and not simply a purely synthetic ending fused to the infinitive.

However, in other branches of Core Romance, mesoclisis is unattested even at their earliest stages, suggesting that full fusion must have taken place in most of the prestige Core Romance varieties by the 10th or 11th centuries (noting that the very earliest texts may have just failed to include mesoclisis by chance).

So, in terms of your sequence, and making some educated guesses as to when Sardinian and Balkan Romance split off:

Steps 1 & 2 likely occurred at some point no earlier than the 6th century, at which point the Balkan Romance languages had been cut off by several invasions in the Migration Period, and Sardinia conquered by the Vandals and later the Byzantines, who were at that point using Greek for their administration.

Steps 3 & 4 occurred in most of Core Romance by the 10th or 11th centuries, with step 4 occurring earlier, and more widely.