In this paper, Naudé argues that the differences between Early Biblical Hebrew (EBH) and Late Biblical Hebrew (LBH) are not entirely explained by a linear chronological model of language change. Rather, he seeks to make a finer distinction between language change and language diffusion.
The notions “language” and “dialect” tend to be described socio-politically, but are ill defined empirically and difficult to study formally. Strictly speaking, it is the individual speaker and their idiolect that can be studied and quantified. Each individual constructs their own grammar (note that Naudé is arguing from a generative framework) during the period of language acquisition. The model of the parent provides some basis, but inevitably the child will end up with a grammar that is not identical to that of the parent. Further, at a certain point when input differs from their accepted grammar they will begin to form additional grammars rather than revise their constructed grammar. Thus most speakers have multiple grammars for different styles, registers, local dialects, etc.
Therefore, in Naudé’s scheme, language change properly occurs at the level of the individual rather than being a function of the language as a whole. Many of these changes will die with the individual, but others will spread to other speakers based on many factors. Naudé labels this process diffusion. Diffusion naturally leads to variation within a population as the old and new features exist side by side. The process seems to be gradual at first until a critical mass of speakers is reached, at which point there is a rapid increase in the spread of the feature.
Thus, the differences between EBH and LBH are not evidence of language change, but of the broader diffusion of features that were already existent within EBH. Several different grammars can be found within LBH: P, Ezekiel, Ezra, Nehemiah, Chronicles, etc. Using Polzin’s 19 features of LBH (See Robert Polzin, Late Biblical Hebrew: Toward an Historical Typology of Biblical Hebrew Prose, Polzin lists 13 changes not attributable to Aramaic influence, A1-A13, and 6 changes that are attributable to Aramaic influence, B1-B6). Naudé finds that P contains five of these changes (A2, A6, A7, A9, A11). Ezekiel shows the diffusion of three of these five changes (A2, A7, A9 – assuming that P represents the parent language of Ezekiel) and four new changes (A4, A5, B1, B2). Ezra reflect the diffusion of the other two changes in P (A6, A11), three of the changes in Ezekiel (A4, A5, B2), and four new changes (A1, A8, A12, B6). In contrast, Chronicles only has one new change (B3). This suggest that none of these specific grammars should be regarded as the transition between EBH and LBH, but rather the process of diffusion is continuous.