The dissertation presents the project of etymological dictionary of Romance languages, Dictionnaire étymologique roman – DÉRom, and compares its methodology with its predecessor, Romanisches Etymologisches Wörterbuch – REW, while also addressing the issue of written, spoken and reconstructed language on the example of Latin, the ancestor language of Romance languages.
The development of the word for 'egg' is studied in each of the Romance languages, as well as in the form of a dictionary entry as it will be published in DÉRom. The established reconstructed form, */'ɔβ-u/, is compared with Latin ōvum; in the case of the divergent word-initial vocalism the existing explanations are critically examined, while highliting the most likely option, which presupposes the drop of the intervocalic consonant, the shortening of the vowel in hiatus and the subsequent paradigmatic leveling in which the intervocalic consonant was inserted back into its previous position.
The word is then considered from the Indo-European perspective, starting with the individual branches and followed by the discussion of the problems introduced by the indicated reconstructions, especially in relation to Sievers' Law. Three proposals of the reconstructed form are considered, determining that the most suitable is the oldest proposal, a thematised derivation with vr̥ddhi from the word for 'bird'.
On this basis, an original acrostatic paradigm of type I (*h2ou̯-i-/h2eu̯-i-) is set for 'bird', but the actually attested words can only be explained by a later transition to a partially proterokinetic and then a fully proterokinetic paradigm, in the case of the latter motivated by the homonym 'sheep' (these two words that were initially phonetically and paradigmatically identical were thus differentiated by adopting different accentual paradigms). In Indo-Iranian a secondary paradigm developed, based on weak stem of the proterokinetic paradigm.