What is the correct etymology of ignōscō "pardon"?

Question

The verb ignōscō, with the meaning "pardon, forgive", is explained in some sources as coming from the negative prefix in- and (g)nōscō. For example, Lewis and Short says "lit., not to wish to know, not to search into; hence, with esp. reference to a fault or crime, to pardon". Page 6 of this handout also explains it this way.

But the negative prefix in- is not usually applied to finite verbs. I found a passage by C. Knapp from 1917 that addresses this topic:

there are extremely few finite verb-forms which show the negative prefix in. One finds it hard to recall any example save indecet. The difficulty set up by this fact of Latin word-formation is so great that various more or less complicated explanations of the etymology of ignosco have been suggested. However, there is a tendency at present to return to the explanation that the word was made up of the negative prefix in and (g)nosco. It is pointed out that as an intermediary to such a form as indecet lies the adjective indecens [...] Back of ignosco may lie ignotus, as the intermediary form.

("The Teaching of Vergil in Secondary Schools", October 8 1917, in The Classical Weekly, Volume XI, New York, 1917-1918, p. 12)

Can anyone tell me about more recent scholarship on this topic?

Answer 1

In a comment, Alex B. referred to the 2001 Dictionnaire étymologique de la langue latine by Alfred Ernout and Antoine Meillet, which contains the following passage:

Les grammairiens latins voyaient dans ignōscere un composé avec le préfixe privatif in-; cf. la glose ignoscere : non noscere, Loewe, Prodromus 409, et Thes. gloss. emend. s. u. ignōscō. Mais la négation in- ne s'emploie pas devant un verbe, cf. plus loin sous in-, et c'est arbitrairement que l'on suppose (encore récemmant Immisch, Glotta, 19, 16-24) que ignōscō aurait été créé sur ignōscēns (sīs, fuās) ; l'exemple de indecet formé sur indecēns n'est pas probant, car indecet est rare et d'apparaît pas avant Pline le Jeune, tandis que ignōscō est ancient et usuel. D'autre part, le participe présent n'est pas d'un usage tellement fréquent ; et, sémantiquement, il marque un état qui dure. Ignōscēns sīs ne saurait donc se dire dans le sens de « pardonne » : quand Térence dit animus ignoscentior, Heaut. 635, il l'emploie comme adjectif pour marquer une disposition de l'esprit avec le sens de « porté au pardon ». Le grec a également avec le même sens un autre préverbe : συγγιγνώσκω, συγγνώμη; ceci suggère un développement de sens tel que « s'accorder avec, sympathiser ». Wackernagel, Mél. Danielsson, p. 383 sqq., a rapproché avec ingéniosité ignosco de skr. anujñā « permettre ». Ignōscō renfermerait un préverbe in- comparable au skr. anu-, que M. Leumann propose de voir aussi dans inueniō, inuidiō ; mais l'existence de ce préverbe dans ces mots est des plus douteuses, et le passage de *enu a in- malaisé à admettre. On a proposé *in-gnōscō, mais le sens n'est pas expliqué par là.

(p. 308)

My translation (with help from jlliagre on French SE for the translation of « porté au pardon »):

The Latin grammarians saw in ignōscere a compound with the privative prefix in-; cf. the gloss ignoscere : non noscere, Loewe, Prodromus 409, and Thes. gloss. emend. s. u. ignōscō. But the negation in- is not used before a verb, cf. below under in-, and it is arbitrary to suppose (as recently as Immisch, Glotta, 19, 16-24) that ignōscō would have been created based on ignōscēns (sīs, fuās) ; the example of indecet formed based on indecēns is not compelling, because indecet is rare and does not appear before Pliny the Younger, while ignōscō is old and usual. Furthermore, the usage of the present participle is not so frequent, and semantically, it marks a continuing state. Ignōscēns sīs cannot therefore mean "pardon": when Terence says animus ignoscentior, Heaut. 635, he uses it as an adjective to mark a disposition of the spirit with the sense of "inclined to give pardon". Greek has likewise with the same sense another prefix: συγγιγνώσκω, συγγνώμη; which suggests a development of sense such as "agree with, sympathize". Wackernagel, Mél. Danielsson, p. 383 sqq., ingeniously compared ignōscō and Sanskrit anujñā "permit". Ignōscō would contain a prefix in- comparable to the Sanskrit anu-, that M. Leumann proposes can be seen also in inueniō, inuidiō ; but the existence of this prefix in these words is of the greatest doubt, and the passage from *enu to in- not easy to admit. *In-gnōscō has been proposed, but this does not explain the meaning.

Answer 2

As for more recent scholarship on this topic, I recommend you to take a look at the following monograph by G. Haverling, who is THE expert on -sco verbs.

HAVERLING, Gerd (2000). On sco-Verbs, Prefixes and Semantic Functions. A Study in the Development of Prefixed and Unprefixed Verbs from Early to Late Latin. Studia Graeca et Latina Gothoburgensia, 64. Göteborg: Acta Universitatis Gothoburgensis.

Interestingly (or perhaps I should rather say "Quite surprisingly"), this scholar appears to claim that the prefix of ignosco is not the negative prefix IN- (as has traditionally been assumed) but rather the lative prefix IN-. This opinion is shared by Prof. Benjamín García-Hernández (another important Latinist scholar who has worked on the topic of Latin preverbs quite extensively: e.g., García-Hernández 1989). However, as pointed out in his review of Haverling's (2000) book/page 131, he is not convinced by Haverling's proposal of deriving the meaning of forgiving from the ingressive meaning of beginning to get to know. Unfortunately, he does not offer any alternative proposal.

Pending a convincing explanation, I'd buy Knapp's point quoted in your post. After all, on the basis that the stative verb ignoro is formed from the prefixed adjective ignarus (<in+gnarus), why not to assume that an analogical process could derive the verb ignosco from ignotus via backformation (along with the polysemous change above taken from Lewis and Short: "lit., not to wish to know, not to search into; hence, with esp. reference to a fault or crime, to pardon")?