Substrata Residue, Linguistic Reconstruction, and Linking: Methodological Premises, and the Case History of Palaeo-Sardinian

This paper demonstrates that, within substratal research, prior to undertaking any comparative endeavour, it is necessary to conduct a thorough distributional analysis of the morphological regularities displayed by the language under consideration, so as to determine the phonological rules governing diachronic changes, which leads to establishing the typology of the substratal language. The application of this rigorous methodology to Palaeo-Sardinian toponymic material makes it possible to recognize the primitive agglutinative typology, and thereby to precise its relation to Palaeo-Basque. After having highlighted some flaws and weaknesses of prior reconstructions, the author first describes the benefits stemming from a systematic segmentation of nearly 1000 microtoponyms of Central Sardinia, which display clear morphological regularities, and restores the underlying phonological system, as well as some of the most distinctive evolutionary laws (e.g., it is argued that the structure of most reconstructed roots can be boiled down to a single syllable template CVC, as /d-u-r/, /d-o-n/; this helps to establish some phonetic laws, as /d/ > /l/ in dur > lur, don > loh, etc.). Finally, a detailed confrontation of Palaeo-Sardinian with reconstructed morphological and phonological systems of Palaeo-Basque evince a vast array of striking correspondances which are due, most probably, to the prehistoric split of Pre-Proto-Basque into Proto-Basque and Palaeo-Sardinian branches in the late Mesolithic / early Neolithic age. The paper provides a new Stammbaum model to account for this split