The central topic of the paper is clitic-verb non-adjacency (interpolation) in Classical and early Modern European Portuguese (EP). Throughout that period, the negative marker não was the only expression likely to disrupt the continuity of clitic-verb sequences. The aims of the study are twofold. First, previous assumptions on the syntax of this pattern are matched against data retrieved from the texts stored in the Biblioteca Nacional Digital. The present analysis demonstrates, first, that clitic-verb adjacency and non-adjacency had divergent distributions depending on whether they occurred in obligatory proclisis contexts or in X-V-cl/X-cl-V variation ones. Whereas interpolation was the default choice in the former, clitic-verb adjacency was prevailing in the latter in the period under discussion. The second aim is to account for the overrepresentation, underpinned by corpus data, of 3rd person accusative pronouns in clauses with interpolation. In previous works, the origin of clitic-verb non-adjacency was couched either in semantic terms (the contribution of não to the meaning of the VP) or in terms of clitic movement and increasing constraints on the status of Neg (Neg and clitic are both assumed to be affixes). In the present paper, interpolation is shown to have had an interface nature, with phonological factors playing a central role. The discontinuity of clitic-Verb sequences is thought to have helped speakers to get rid of the affixal attachment of the o, a, os, as series to the preceding non-verbal sound material (nasal diphthongs in não, quem, ninguém, etc, coercing preverbal pronouns into taking a nasal onset quem no, não no, etc.). As a consequence, in present day standard EP, clitic-specific allomorphy has become earmarked for enclisis.