¹ I do not know how T. would account for the fact that a length contrast in consonants still exists in the present-day dialects of Cyprus, the Dodecanese, Chios, some of the Cyclades, Kimi (Euboia), and South Italy: see B. Newton, The Generative Interpretation of Dialect, pp. 89–93. Similarly his hypothesis of early merger of υ and ι is at least disconfirmed by the fact that υ is frequently represented by [u] in various modern dialects closely related to the pre-independence dialect of Athens (!) (Newton, op. cit., 20–3). Although T.s method depends crucially on knowledge of later historical developments, his bibliography contains nothing on Modern Greek except the textbooks of Hatzidakis, Pernot, and Thumb.

² Some of these false writings are probably not phonemic-graphemic confusions, but narrowphonetic or morphophonemic spelling (though T. does not note this); even when these are ignored, though, the number of instances remaining is still greater than for many confusions which T. accepts as evidence of phonemic change.

³ T. is surprisingly silent on the possibility of geographical divisions within Attic; nor does he give the provenance of the documents containing the various misspellings.