What did the Romans consider the "basic" form of a verb?

Question

Many of us are used to using the (active present) infinitive form of a verb as a "label" or "basic form" or "representative" of the verb. By this I refer to uses like dictionary entries or grammatical statements similar to "the active indicative perfect second person singular of amare is amasti". Some Latin dictionaries use amo instead of amare for these purposes. There was a separate question on this issue.

What did the Romans themselves consider the basic form of a verb? If it was inconsistent, which methods did people use to refer to verbs in grammatical context? I think such a "label form of a verb" was used mainly by grammarians and etymologists — or other writers temporarily acting as such.

Answer 1

First person singular (laudo) appears to be most common

Marcus Terentius Varro (116-27 BC) wrote De Lingua Latina, which survives in partial, corrupted form, but which provides valuable testimony on this matter.

In Book 6, chapter 5, he speaks of "derived words" and uses the first-person present active indicative form as the "basic form":

Primigenia dicuntur verba ut lego, scribo, sto, sedeo et cetera, quae non sunt ab alio quo verbo, sed suas habent radices. Contra verba declinata sunt, quae ab alio quo oriuntur, ut ab lego legis, legit, legam et sic indidem hinc permulta. Quare si quis primigeniorum verborum origines ostenderit, si ea mille sunt, quingentum milium simplicium verborum causas aperuerit una; sin nullius, tamen qui ab his reliqua orta ostenderit, satis dixerit de originibus verborum, cum unde nata sint, principia erunt pauca, quae inde nata sint, innumerabilia.

Shortly before, he also speaks of deriving nouns from verbs, and again uses this form as the base case:

Cum verborum declinatuum genera sint quattuor, unum quod tempora adsignificat neque habet casus, ut ab lego leges, lege; alterum quod casus habet neque tempora adsignificat, ut ab lego lectio et lector; tertium quod habet utrunque et tempora et casus, ut ab lego legens, lecturus; quartum quod neutrum habet, ut ab lego lecte ac lectissime.

Another important grammatical work is the Institutio Oratoria of Quintilian (AD 35-100). There is a case where he seems to default to the third-person singular:

Neque has modo noverit mutationes, quas adferunt declinatio aut praepositio, ut "secat secuit", "cadit excidit", "caedit excidit", "calcat exculcat" (et fit a "lavando" "lotus" et inde rursus "inlutus", et mille alia), sed et quae rectis quoque casibus aetate transierunt. (I, 4)

Shortly afterwards, though, he uses the first-person as a kind of default, contrasting it with impersonal verbs:

Quaedam [verba] etiam mutantur, ut "fero" in praeterito, quaedam tertiae demum personae figura dicuntur, ut "licet" "piget". (ibid)

This second form, in accord with Varro, seems to predominate. The next section uses the first-person in an unambiguous manner:

nam duplex [verborum] intellectus est, alter qui omnia per quae sermo nectitur significat, ut apud Horatium: "verbaque provisam rem non invita sequentur"; alter in quo est una pars orationis: "lego" "scribo" (I, 5)

While this certainly isn't definitive proof (nor, in my mind, is it likely to find any), I believe it's at least enough attestation that referring to a base verb by its first-person form is acceptable and precedented.

...but infinitive (laudare) has attestations as well

Although the infinitive does not appear as common, it can be used as well, as demonstrated in the following passage from Quintilian:

Quo modo autem "quire" et "urgere" vel in praeterita patiendi modo vel in participia transibunt? (I, 9)