This article aims to provide a detailed exposition on the treatment of the speech-sound and letters in ancient Latin grammatical tradition and thereby to show the rationale of the mode of their explanation. In ancient Latin grammars, a letter is recognised as an entity which has three properties, i.e. name, figure, and sound. Donatus`s definitions of vox and littera show the close relation between the two concepts, of which the crucial link is ``meaning`` as is clearly explained by Priscian. Moreover, the aural aspect of the speech-sound expressed in the definition of speech-sound and the etymological explanation of littera as legendi iter need to be understood against the backdrop of the oral/aural culture of the ancient period. The emphasis placed on the oral aspect of the letter in the ancient Latin grammars reflects the function and the aim of the grammatical education in the period, namely the interpretation of the literature and speaking correctly. Accordingly, the treatment of the speech-sound and the letter was developed with this literal and rhetorical orientation. The phonological treatment of the ancient Latin grammars needs to be considered and evaluated within this context.