French historical development: 1p singular

[So Haleza Grise]

So many questions about French in this forum! Oh well, here's another one.

I can see how it's possible to get sound changes to from (say) credis to crois or from facis to fais. Similarly there's no great mystery to me how credit ends up as croit and facit as fait.

What I do wonder about is what happened to the 1p singular forms of these verbs. What were the reflexes of credo and facio? At some point they were obviously replaced by the 2p singular forms - I am not sure if what motivated an orthographic replacement. After all, fais and fait merged in sound but have been kept orthographically distinct. And we still have an orthographically distinct form for chante <-- canto in the first conjugation.

How did it happen?

[Linguoboy]

Isn’t the development of facio > fais parallel to pacem > paix?

[Ares Land]

So many questions about French in this forum! Oh well, here's another one.

I can see how it's possible to get sound changes to from (say) credis to crois or from facis to fais. Similarly there's no great mystery to me how credit ends up as croit and facit as fait.

What I do wonder about is what happened to the 1p singular forms of these verbs. What were the reflexes of credo and facio? At some point they were obviously replaced by the 2p singular forms - I am not sure if what motivated an orthographic replacement. After all, fais and fait merged in sound but have been kept orthographically distinct. And we still have an orthographically distinct form for chante <-- canto in the first conjugation.

How did it happen?

Your first question is fairly easy to answer: crei and faz. In other words, the -s wasn't there in the first place (well, except that z is [ts] but it's from Latin ci in that case). It wasn't in suis either.

As for your second answer, well, sigh. The evolution of Romance verbs is just awful, paradigms were levelled by analogy for no clear reason.

Old French had the perfectly reasonable sui or soi, and yet added an s anyway. For some reason an -s was added. For that matter Italian had *so > sono for no clear reason, and generalized -iamo, of all possible endings in the first person plural, whereas Italian verbs already had very sensible endings. Go figure. Apparently the first person is highly susceptible to being reworked.

We can try to provide an explanation anyway:

As for faire, OFr had faz, fais, fet, faimes, faites, font. Root alternation was very common in OFr but was mostly regularized in MFr.
So faz -- that was something like [fats] was regularized by analogy with the more common fais (which was either [fajs] or [fɛs]). That sounds reasonable.

And then the ending -s was generalized to all verbs in the third conjugation class.

In brief: OFr conjugation was just awful and, as is common in Romance, was regularized in an expedient but counter-intuitive in MFr.

[Kuchigakatai]

Latin inchoatives in -scō merged into their base verbs in Old French (so flōrēscere and flōrēre became one verb), so OF ended up with verbs that had forms descending from the -scō inchoative in the present indicative and present subjunctive, and forms from the base verb in the other tenses. This pattern expanded to some Old French -ir verbs (and as time would pass, also many more verbs).

These pres. indic. forms descending from ancient inchoatives had 1SG -s and 2SG -s:
fīniō fīnīs > *fīnīscō *fīnīscis > je finis tu finis (inf. fīnīre > fenir)
cognōscō cognōscis > je conois tu conois (inf. cognōscere > conoistre)
crēscō crēscis > je crois tu crois (inf. crēscere > croistre)
nāscor nāsceris > analog. *[ˈnasko nastses] > je nais tu nais (inf. nāscī > analog. *nāscere > naistre)

To these, we add Latin verbs whose stems ended in a consonant that became [s] in French through regular sound changes or by a slight analogy of [ts] > [s]:
cōnsuō cōnsuis > OF je *cos tu *cos > Late OF je queus tu queus (inf. cōnsuere > cosre > coudre)
exeō exis > je is tu is (inf. exīre > eissir/issir)
jaceō jacēs > analog. je gis tu gis (analogized, expected forms: *giz) (inf. jacēre > analog. *jacīre > gisir)
lūceō lūcet > analog. je luis tu luis (analogized) (inf. lūcēre > analog. *lūcīre/*lūcere > luisir/luire)

These verbs above seem to have paved the way for the expansion of -s in the present indicative 1SG form of almost all -oir, -re and -ir verbs in Middle French.

videō vidēs > *[ˈve:do ˈve:des] > OF je voi tu vois > MF je vois tu vois (inf. vidēre > EOF vedeir [vəˈðejr] > OF veoir > MF voir)
crēdō crēdis > OF je croi tu croi > MF je crois tu crois (inf. crēdere > EOF creidre > OF croire)
faciō facis > OF je fai (rare je faz) tu fais > MF je fais tu fais (inf. facere > faire)
serviō servīs > OF je serf tu sers > MF je sers tu sers (inf. servīre > OF servir)

The only verbs in the -oir/-re/-ir groups that managed to escape were 1) verbs with a stem ending in labial + /r/ 2) verbs with a stem ending in /ʎ/ which analogized to -er verbs while retaining some -ir forms:
cooperiō cooperīs > > OF je covre tu covres > MF je couvre tu couvres
colligō colligis > *[ˈkɔljo ˈkɔles] > OF je cueil [ˈkwɛʎ] tu cuelz [ˈkwɛɫts] (inf. colligere > coillir / analog. cueildre)
> analogized to MF je cueille [ˈkœʎə] tu cueilles

Salīre > saillir is another verb that escaped 1SG -s by analogizing to -er verbs (je saille tu sailles).

Notice that fallere > analog. *fallīre > faillir and bullīre > bouillir present the expected Middle French outcomes:
fallō fallis > [ˈfal:jo ˈfal:es] > OF je fail [faʎ] tu falz [faɫts] > MF je faux tu faux [fos]
bulliō bullis > OF je boil tu bols > MF je boux tu boux [bus]
([faɫts] > [fawts] > [faws] > [fos], [bols] > [bows] > [bus])

For that matter Italian had *so > sono for no clear reason, and generalized -iamo, of all possible endings in the first person plural, whereas Italian verbs already had very sensible endings. Go figure.

I personally somewhat buy the idea that that 3PL sunt [son] got an [o] attached as an analogy to the 3PL -unt of -ere/īre verbs (dūcunt/capiunt/audiunt), becoming *sunt-unt [ˈsono]. And Florentine at the time had so many of these same verbs that were pronounced the same in the 1SG and 3PL forms (just like Romanian today), as in dūcō/dūcunt *[ˈduːko] or audiō/audiunt [ˈawd(j)o], that the copula expanded its new sunt-unt [ˈsono] to 1SG sum, i.e. sum-unt [ˈsono]. This 3PL -[no] ending then also expanded throughout all 3PL forms.

I like it because this would explain the geminate -nn- in a few verbs like habent > *[an] > hanno [ˈan:o], vādunt > *[van] > vanno, faciunt > *[fan] > fanno, dan > danno, stant > stanno (cf. Spanish habent>han, van, dan, están).
That is, Florentine hanno would etymologically come from habent-(su)nt-unt [an-n-o].

Further evidence for the tight relationship of 1SG and 3PL in pre-Italian is the expansion of 1SG -go to 3PL, something not seen in most other Western/Italo-Romance varieties: 3PL pōnunt [ˈpo:no] > *[pon-go] > then pongono, vident > *vid-unt [ˈve:do] > *[ved-go] > then veggono. Also the stem in the now old-fashioned 3PL jacent > giacciono, with a geminate -cc- that has been analogized from 1SG jaciō > giaccio.


The replacement of Old Florentine amamo/tenemo/udimo by the subjunctive -eāmus/-iāmus > -iamo, giving amiamo/teniamo/udiamo, is very head-scratching though, I agree. Why did they all grab the subjunctive 1PL while retaining the inherited indicative 2PL (amate/tenete/udite)?!?? Even -are verbs did it! WTF?

[Ares Land]

Neat! I didn't think of the inchoative... Thanks!

For Italian third person, the explanation is apparently as follows:
nt was systematically dropped, so, as you say 1SG and 3PL were often identical.

But essere had 1SG *son. *son was reinterprated as a verb root, so 1st person -o was added. The 3PL became sono by analogy, and -no was reinterpreted as a clitic third person plural marker.
As for the geminate n, well, either it was triggered by the etymological nt, or n was lengthened due to ha carrying stress (so the syllable had to be heavy).
Still, I stand by my original 'go figure'

As for -iamo, the official answer seems to be, well, WTF??!

[Kuchigakatai]

Oh, right, 1SG -o. I forgot that's the common (practically official) explanation. Nevertheless I find it neat the expansion of that to 3PL sunt [son] was supported by every other -ere/īre verb ending in [o] too (currunt [ˈkor:o]).

As for the geminate n, well, either it was triggered by the etymological nt, or n was lengthened due to ha carrying stress (so the syllable had to be heavy).

My only issue with the second option is that these verbs had very reduced forms in pre-Italian due to being unstressed. By regular sound changes habent and vādunt would've ended up as *habono and *vadono.

As for -iamo, the official answer seems to be, well, WTF??!

Ironic too, considering French went in the opposite direction, replacing its inherited present subjunctive -iens, as in jungāmus > joigniens [ʒoˈɲjɛ̃ns] (from -eāmus/iāmus > iens; -āmus > reinterpreted as *[ja:mos] due to pres. subj. often having a "palatal stem" and analog. to -eāmus/iāmus) by the indicative -ons (joignons).

Then the same goes for the imperfect subjunctive, where inherited 1PL -īvissēmus > -issiens and 2PL -īvissētis > -issoiz, attested in Old French, were later on analogized to the present subjunctive (itself modelled after the present indicative), so -issions and -issiez. Italian of course decides to go both ways here, but taking the opposite path of its present indicative, keeping its inherited 1PL (amāvissēmus > amāssēmus > amassimo [aˈmas:imo]) and merging its 2PL with the preterite (amaste).

Seriously, it's surprising that Old French would remodel so much stuff about its verbal conjugation off the Latin -ēre/ere/īre endings, instead of the more common and regular -āre ones. The expected -āre present subjunctive endings -ēmus > *-eins/ins (parler/doner: *parleins/*doignins) and -ētis > *-oiz/iz (*parloiz/doigniz) are nowhere to be found.

It's such a mess.

[Zju]

I personally somewhat buy the idea that that 3PL sunt [son] got an [o] attached as an analogy to the 3PL -unt of -ere/īre verbs (dūcunt/capiunt/audiunt), becoming *sunt-unt [ˈsono]. And Florentine at the time had so many of these same verbs that were pronounced the same in the 1SG and 3PL forms (just like Romanian today), as in dūcō/dūcunt *[ˈduːko] or audiō/audiunt [ˈawd(j)o], that the copula expanded its new sunt-unt [ˈsono] to 1SG sum, i.e. sum-unt [ˈsono]. This 3PL -[no] ending then also expanded throughout all 3PL forms.

I like it because this would explain the geminate -nn- in a few verbs like habent > *[an] > hanno [ˈan:o], vādunt > *[van] > vanno, faciunt > *[fan] > fanno, dan > danno, stant > stanno (cf. Spanish habent>han, van, dan, están).
That is, Florentine hanno would etymologically come from habent-(su)nt-unt [an-n-o].

Why didn't *van *fan *dan become *vano *fano *dano during the first stage, just like *son > sono?

[Kuchigakatai]

Why didn't *van *fan *dan become *vano *fano *dano during the first stage, just like *son > sono?

Because these highly common verbs ended up with different 1SG and 3PL forms, having either reduced both the 1SG and 3PL forms or only the 3PL. In contrast, the copula and most -ere/īre verbs ended up with identical forms (a pattern later on expanded to -ēre verbs as well). 3PL sunt [son] became *sunt-unt (or *sunt-ō if you prefer) [sono] because 1SG sum [son] also became *sum-ō [sono], partially as analogy with the -[o] of all 1SG verbs, partially as analogy with the identity of 1SG and 3PL forms of -ere/īre verbs (and the copula *esse-re was one of them).

This means that at some point in pre-Italian there existed a split between:
* 1) a few highly common verbs and the large number of -āre verbs, a productive category, which distinguished 1SG and 3PL,
* 2) the copula *[ˈɛs:ere]/*[sono] and all the -ere/īre verbs, which were not that productive but were very common, which didn't distinguish 1SG and 3PL.

vādō [vao]>[vo] vādunt [van]
faciō [fattʃjo fajo]>[fattʃo fo] faciunt [fan]
habeō [ajo]>[o] habent [an]
dō > [do] dant > [dan]
stō [esˈto] stant [esˈtan]
amō [ˈa:mo] amant [ˈa:ma] (amāre)

sum [son] sunt [son] > *sum-ō [son-o] *sunt-unt/*sunt-ō [son-o]
clūdō [ˈklu:do] clūdunt [ˈklu:do] (clūdere)
*fīnīscō [fiˈnisko] *fīnīscunt [fiˈnisko] (fīnīre)

The forms on the right would then surface in Old Italian as van-no, fan-no, han-no, dan-no, stan-no, ama-no on the one hand, and as son-o, cludo-no, finisco-no on the other. (-ēre verbs pattern with -ere/īre verbs, see movēre: movent > *mov-unt [ˈmwɔvo] > Italian muovo-no.)

[So Haleza Grise]

As for your second answer, well, sigh. The evolution of Romance verbs is just awful, paradigms were levelled by analogy for no clear reason.
[...]
In brief: OFr conjugation was just awful and, as is common in Romance, was regularized in an expedient but counter-intuitive in MFr.

Thank you! I should have thought this was the reason. It's nice to have it all laid out like this. Such a mess!