This study focuses on accounting for allomorphy in Latin case/number inflection. It attributes essentially all of it to the influence of adjacent features of standard segmental phonology on morphemes expressing case assignment. Indeed, other languages lead linguists to expect that allomorphy at a stem-suffix border can depend on any feature of a final vowel: ±Low, ±High, ±Round, ±Back, ±Consonantal or ±Syllabic. Empirically, it turns out that no two Latin stem-final vowels induce identical allomorphy in case/number suffixes, nor the same allomorphy as final consonants. Moreover, some (not much) phonologically conditioned allomorphy is phonetically opaque.
These two factors have led traditional scholarship to conclude that each stem-final segment should define a separate classification of noun suffixation or “declension”. As there are basically six types of final segment (a, o, u, i, e and consonants), each then gives rise to a different declension (stems with final i are often put in some other group or considered irregular), thus creating five (or six) declensions. Tradition then goes on to analyse stem-final vowels not as part of stems but as some kind of separate morphemes called “thematic vowels” (with no role in either syntax or phonology).
This essay argues rather that an unflinching modern and formal approach to inflectional allomorphy, exactly analogous to using phonology to reduce regular English plurals to a single lexical form, succeeds in sweeping away the sandcastle of Latin declensions and better captures the actual descriptive generalizations that account for Latin case inflection.