A study on the dialectology of Vulgar Latin vocalic mergers: the interaction between vowel quality, syncope and accent Béla Adamik Eötvös Loránd University Budapest adamik.bela@btk.elte.hu ABSTRACT This paper contributes to the issue of a potential correlation between the proportion of vocalic confusions under the accent measured against the proportion in unaccented syllables and the intensity of the accent itself, as supposed by József Herman in 1965. Accordingly, where there are relatively many mistakes in accented syllables, i.e. where we find a higher proportion of misspellings as for the vowel mergers in stressed syllables, there seems to be a relatively lower intensity of stress; and, conversely: where we find relatively few mistakes in stressed syllables, i.e. where we record a lower proportion of misspellings of the vowel mergers in stressed syllables, there the stress must have been of higher intensity. This correlation theory and the underlying findings of Herman were criticized and rejected by J. N. Adams in 2007. By adding the phenomenon of syncope to the equation, and looking at data sets from the “Computerized Historical Linguistic Database of Latin Inscriptions of the Imperial Age” (http://lldb.elte.hu/), recorded from a selection of Roman provinces, this survey will prove that the correlation suggested by Herman indeed existed and it was in effect throughout the whole Late Latin period. 1. Introduction 1.1. In his study on territorial differences in the phonology of Latin during the Empire, published first in 1965 then again in 1990, in the section ‘The problem of stress accent’ (Herman 1965=1990: 22-24), József Herman discussed the territorial distribution of the orthographic confusions 1 between E and I and between O and U corresponding to the well known Vulgar Latin vowel mergers of long ē and short i as a close ẹ, long ō and short u as a close ọ in stressed syllables, and to the same mergers but extended also to short e and short o respectively in unstressed syllables. He analyzed the late (i.e. 5 th 1 In this survey for denoting the various types of misspellings in inscriptions I use the code- system of the Computerized Historical Linguistic Database of Latin Inscriptions of the Imperial Age (cf. http://lldb.elte.hu/ and accordingly abbreviated as Database or LLDB hereafter); as for the format of the codes, the sign “>” is to be interpreted as “represented in the inscriptional text as”, e.g. “é: > I” means “a Classical Latin stressed long e is represented in the text by a letter I”.