Could a child be described as 'stout-hearted, determined, stubborn'? Could νήπιον αὔτως mean 'yet still a child' marking a contrast with 'stout-hearted'? Babies and toddlers can be quite determined. This one used to eat 'only marrow and the rich fat of sheep' (Il.22.501). He will refused to go to Hector's arms (Il.6.467-470) and put up quite a fuss, but there fear seems to be the point (ἀτυχθείς, ταρβήσας, δεινόν). Il.22.486 has πάις δ’ ἔτι νήπιον αὔτως, a youth, yet still in infancy', which recalls 6.400 but with the important addition of δέ and ἔτι.
The transmitted adjective is regularly glossed by ἁπαλόφρων, which otherwise is only in Marcus Argentarius (Anth.7.403).
The transmitted adjective is otherwise only in Q.S.13.122-123, which shares other items of vocabulary with the Iliadic passage: Οἰμωγὴ δ’ ἀταλάφρονας ἔκβαλεν ὕπνου | νηπιάχους τῶν οὔ πω ἐπίστατο κήδεα θυμός. For νηπιάχους (Il.6.408), cf. GVI 741 (2nd. c. CE or later).
The form expected from ἀταλός would be ἀταλ-ό-φρων, which is attested epigraphically and as a personal name.
Leumann proposed that ἀταλός is a ghost word that arose in the epic tradition when (de-adjectival) *ἀταλαφρονέων was misunderstood as ἀταλὰ φρονέων (cf. Il.18.567 masc.nom.pl., Hes.Th.989 masc.acc.sg., hHom.2.24 fem.nom.sg.). As M.L. West noted, other members of the word group (esp. ἀτιτάλλω and, at Gortyn, ἀτιτάλτᾱς) imply great antiquity for ἀταλός, whose neut.pl.super.adv. appears as early as the Dipylon Oenochoe (closely comparable to the neut.pl.adv. in Hom. Hes. and hHom.). Beekes wrote:
"This [*ἀταλαφρονέων ] is based on ἀταλάφρων, and this again is the negation of ταλάφρων. Although it explains the compositional vowel α, the whole analysis is too complicated.".
Perhaps, then, it is simpler to address the question of "the compositional vowel α" in isolation from ἀταλός and its derivatives. There seems to be a word group based on a real stem ἀταλο-. The adjective ἀταλάφρων should be investigated in isolation: (1) is it the negative of ταλάφρων, misunderstood by later poets, namely Quintus Smyrnaeus? (2) Is it the product of misanalysis of the Iliadic line? (3) Or should a correction to ἀταλόφρονα be implemented?
(1) For alpha privative as the third element of a compound, cf. ἀνιπτόπους in the Iliad, ἀταλαίπωρος said to be first in Aeschylus and Thucydides, ἀωρόνυκτος in Aeschylus, and ἀπειρομάχᾱς in Pindar. It is possible that it was misunderstood within the Homeric tradition, but not such that it gave rise to ἀταλός.
(3) Later poets certainly saw ἀταλάφρονα in Il.6.400 and understood it as 'tender-minded'. By contrast apart from the epigraphy mentioned above, ἀταλόφρονα has left its mark in Hsch. (glossed by νήπια [neut.pl.!], ἁπαλόφρονα), in the scholia to Hes.Op.130 ἀτάλλων (ὁ μετ’ ἐπιμελείας τρεφόμενος, καὶ ὡς ἂν νηπιάζων ἀταλλόφρων (sic),...), and in the Encomium in sanctum Isacium et Dalmatum 29, an eighth-/ninth-century work by Michael Syncellus. Therefore it seems unlikely that ἀταλόφρονα had the prestige of a recherché Homeric epithet.
(θ)εἱλόπεδον (Od.7.123) is an intriguing question involving similar phenomena. See Dubois RPh. 71 (1997): 162. A non-verse inscription (not dated) from Caria turned up θιλοπ[, which has been taken as evidence for the currency of θειλόπεδον in its own right, not as a misunderstanding of εἱλοπεδον.