Classical dialectology has demonstrated that political-administrative and linguistic borders do not always have to coincide. In this paper we look into neighbouring Basque varieties, spoken on both sides of the political border, namely the guipuscoan and navarrese (Spain) and labourdin (France) dialects following the first classification of Basque dialects (Bonaparte, 1863). For that purpose, we take three grammar properties into account: a) the dummy verb egin ‘to do’ and the morpheme ba- for emphasis; b) the particles ahal and al; and, c) the non-adjacency of the wh-word and finite verb in embedded contexts.
Our corpus is composed not only of texts from the second part of the XXth century but also of interviews carried out according to the standard criteria of dialectology (Chambers & Trudgill [1998] 2004).
The first characteristic deals with the syntactic means to mark emphasis on the verb, namely the dummy egin (‘do’) mainly used in western dialects and the morpheme ba- particular of eastern dialects (cf. Euskaltzaindia 2016). Data show that both strategies arise in the Labourdin coast:

(1) Autsi ezta iñen.
break not.aux do.fut
‘It won’t break.’
(2) Baut uste senarra eztuen bate ongi.
cl.aux think husband.abs not.have at.all well
‘I do think that her husband wasn’t well at all.’

     The second property concerns the particle ahal used in all dialects and the particle al attested only in Guipuscoan (Euskaltzaindia 1987; de Rijk 2008). Historically ahal − also pronounced as al − has been productive in the three areas not only in declarative contexts but also in biased questions. However, its current function as a question particle in Guipuscoan has led to separate interpretations of the same utterance:

(3) Etorriko al aiz?
come.fut part aux
‘You’re coming?!’ (Labourdin)
‘Are you coming? (Guipuscoan)

The third property looks into indirect questions. It is an acknowledged fact that the wh-word and the finite verb must be adjacent not only in matrix questions but also in embedded ones (Hualde & Ortiz de Urbina 2003); nevertheless, eastern varieties have developed a more flexible behaviour in indirect questions since other constituents can occur between both them (Monforte 2020).

(4) Ez dakit non ote kazeta utzi dudan.
not know where part journal.abs leave aux.c
‘I don’t know where I might have left the journal.’

     In conclusion, considering the properties mentioned above, we see that the border can be impermeable. The first property suggests that there is a linguistic continuum in the cross-border area covered by the guipuscoan, labourdin and navarrese dialects. As for the second characteristic, although the particle ahal is still in use in the labourdin dialect, the grammaticalization of ahal into al arisen in the guipuscoan dialect has not taken this step forward in the former. Finally, there are other innovations which have not spread across the border such as the third property. Indeed, chronology accounts for the different grade of diffusion since the first property is longer attested than the second and third ones.