One of the possible paths for exploring processes whereby shift evolves from mass bilingualism might be to explore the cases where this is not so. The situations that Ferguson termed 'diglossia' can be used initially to try and shed some light on the problem. How is it that, despite considerable interlinguistic distances (as with the Swiss-German or Greek cases) and a clear distribution of 'high' and 'low' functions, these latter are not abandoned from generation to generation in favour of the 'high' functions? Is this not precisely what happens in many cases of language shift? How can situations that are so superficially similar from the point of view of language shift evolve in such crucially different ways? What if mass bilingualism was a necessary but not a sufficient condition to explain the evolution towards the intergenerational abandonment of own varieties? What if, as Norbert Elias said, "when we are dealing the problem of the need for social evolutions, we have to distinguish clearly and concisely between the statement that B has to necessarily follow A and the statement that A necessarily has to precede B"?(12)

What then makes situations with a hierarchized distribution of functions between linguistic varieties that are structurally distant, appear as stable in some cases whereas, in others, the varieties used in individualised communications tend to be abandoned and replaced by institutionalised communications?(13) Which factors would eventually determine these outcomes – diglossia with varying stability (using Ferguson’s concept) versus shift? Rather than looking at strictly language-based structural divergences, we will probably need to focus on the socio-cognitive representations of speakers on the linguistic varieties at issue and, secondly, on the contexts in which these latter are produced and maintained. Let us clarify that, here, we are not discussing why a given variety is adopted, but rather why the other is abandoned.

Firstly, unlike the situations described by Ferguson, as regards the political context of language shift phenomena, the political powers in question very often hope for the result of language abandonment. In many cases, since the very launch of the process of mass dissemination of the state language (which, for the vast majority of the population, is often first experienced when learning to read and write), the explicit aim is not only to transmit a general language of intercommunication, but also to eliminate any other systems of linguistic communication that differ from the model used by the central, supreme political power. Against this background of subordination and dependence, the population (as it becomes more and more competent in the newly-acquired State language) may choose to transfer this language to their children as a basic variety of socialisation, i. e. as a native variety, thus ending the intergenerational transfer of the group’s own vernacular. The change in habitual guidelines will need a clear, ideological and/or practical justification and legitimation, since we are addressing a behaviour that the community clearly values forming part of. Here, the patriotic discourse of a 'national language', which promotes the notion of a single, general language for all citizens, can be used. Gradually, therefore, as part of a process of asymmetrical dissemination among social and geographic groups, this 'national language' will be adopted as a variety of institutionalized communications; it will then be transferred to individualized communications by a generation that is already competent in the language. This generation will transfer the language as native-speakers to the next, which will now be somewhat unfamiliar with the old vernaculars and will make this variety, received as the formal standard (conveniently adapted to colloquial functions), their own habitual language.

On the contrary, the diglossic distribution of functions usually involves the coexistence of varieties that are perceived as being part of the same 'language'. This is particularly clear with the historical Greek and Arabic cases. Whichever is used, the two varieties have always been seen as undeniably 'Arabic' or 'Greek'. The standard does not tend to cause issues of ethnic identity.

As we said earlier, the varieties are in complementary distribution: the standard variety is never used in informal oral, individualized communication and the vernaculars are never used in written form and rarely in very formal speech. The official standard is consciously learned by generation after generation at schools, whereas vernacular varieties are used in everyday and domestic environments and are the first varieties acquired by individuals. In theory therefore, there does not seem to be room for ethnolinguistic conflict, since the varieties do not symbolize this type of opposition. Therefore, the contrast between varieties does not seem to offer speakers a negative representation leading them to abandon the vernaculars in favour of the standard in informal, everyday communication. In fact, the opposite appears to take place.

Basically, then, the reason for the relative stability in these cases of diglossic distribution lies in the political and cognitive dimension: none of the cases usually analysed are situations of political subordination, such as minoritized communities. The perception of dependence, with its negative undertones, and as a result, self-denigration with the adoption of foreign cultural elements as the main reference for behaviour and values, do not need to occur. It therefore seems obvious that intergenerational shift is not caused by the simple facts of bilingualization and the asymmetrical distribution of functions, but rather by the political and economic context in which the bilingualization occurs and the meanings and representations associated with it by its protagonists. (14)

In many cases, the root of the problem lies in these significant representations of the situation and, specifically, in the evaluations and expectations for each linguistic variety at issue.

The functional equilibrium that would allow intergenerational reproduction of the situation collapses if individuals arrive at the conclusion that, all in all, their children would benefit more from the transfer of their L2, rather than their L1, because they see more advantages in the L2 (often linked to socioeconomic aspects and political and cultural prestige). When overt, formal prestige has greater importance than 'covert' prestige – as it has termed by some authors – individuals may decide to change their child’s language. We therefore need to look at the context - the sociopolitical and economic ecosystem - for the factors that may have led to this decision to abandon the intergenerational transfer of the group’s historical language.